Before the advent of superhet
receivers for the domestic market receivers used a design known
as "TRF" or "Tuned Radio Frequency".
In the earliest sets, which looked more like pieces of scientific apparatus you might have found in a school physics lab, there were often separate tuning dials for each set of coils and each knob had to be set precisely in order to get a decent speaker volume. In some areas close to a local transmitter the separate tuned circuits also provided a degree of interference rejection.
Most TRF sets had dials calibrated from 0 to 180 degrees and a tuning aid chart was often fastened to the set so that users could readily set the dials for best reception of the desired station. Searching for new stations was quite laborious as the dials had to be set to roughly the right settings for anything to be heard. Tuning was therefore quite an art.
The design of the Philips Superinductance receivers was a major step forward because only a single tuning dial was used. All the tuned circuits were coupled together, or ganged, so that a user could tune across the entire waveband without loss of gain and therefore even very weak broadcasts could be heard without difficulty.
The Superinductance design required
extreme accuracy in component manufacture and careful construction
to make the components associated with tuning exactly the same.
In particular, coils needed to be wound so that they were identical
in characteristics otherwise the gain of the set would drop off
across the tuning range. To aid alignment or to adjust individual
sets for precise tuning alignment trimmers were used. In the
Philips 834A these trimmers are tubular ceramic based components
having a very low loss and excellent stability.
Moving on to another feature of a British
or European TRF receiver; the method of achieving sufficient
gain for reception of distant stations.
As sets were already fairly pricey
for several reasons and, now with the extra levy demanded by
Marconi, it became apparent that in order to cut costs designs
should make use of the minimum number of valves.
The reaction control unfortunately had
a serious drawback. As the knob was turned there came a point
where the signal suddenly failed to get louder and a faint high
pitched whistle could be heard. The whistle could also appear
if a set tuned to one station was tuned to a different dial setting.
The problem of "howling" speeded
up the design of the superhet receiver, whose local oscillator
was on a different frequency to the tuned station and therefore
didn't upset your neighbour. Most wireless listeners knew about
the reason for interference from poor adjustment of reaction
controls because, in those days, there were dozens of magazines
published for the listener and radio constructor. These magazines
were extremely popular because they informed the man on the street
how to make a receiver at a fraction of the price of a commercial
set. They also carried readers' letters complaining about interference
from a neighbours set.
The Philips Superinductance design included
measures to obviate howling. The set operated with sufficient
feedback to provide a high gain but was arranged so that the
set never broke into oscillation at the signal frequency. This
was accomplished by careful design connected with valve biasing
and by excellent screening between amplifying stages, hence the
use of the copper screening cans.
Now onto the overhaul of the receiver....
The first step in repairing an early
mains receiver is to examine the mains input circuit. When this
set was designed in 1932 for AC mains use, the domestic voltage
could have been virtually anything so there should be a selector
plug or some similar arrangement for setting the local voltage.
This Philips set was satisfactory in this respect as it was already
configured for the range that included 240 volts.
Any mains equipment if not earthed
will have an AC voltage present on exposed metal parts because
of capacitive leakage although the voltage source will have relatively
high impedance and consequently have a very low current capability.
Touching an exposed metal part of an unearthed receiver will
give one a slight tingling or a rubbery sensation in the fingers.
A mains safety check would result in failure.
The next task is to check any other rubber covered cables. The Philips used a length of such cable between the loudspeaker and the output circuit. Many valve receivers employ a loudspeaker which has an integral output transformer. This means the connecting cable will be carrying HT and must therefore be in good condition. The cable in this example had lost all it flexibility and the rubber insulation was cracked and bare wire was visible. I replaced the cable with modern plastic covered wires as the owner was not worried about maintaining authenticity.
There are now a number of checks that
must be made in the interests of prudence. Two particular capacitors
need to be removed and tested, or better, just removed and replaced.
One is the capacitor connecting the anode of the output valve
to chassis. If this were to go short-circuit the output transformer
would be destroyed. Philips actually connect this capacitor across
the output transformer primary so its failure would not damage
the transformer and I found the capacitor (C14) to be reasonable
in terms of DC leakage, but wildly out in terms of capacitance.
It measured around ten times its marked value of 0.005uF and
would result in lots of top cut, or loss of sharpness in tone.
Roughly speaking... as an example.
If the anode of the LF amplifier carries 100 volts and the input
impedance of the output valve is a megohm then a coupling capacitor
leak of a megohm will result in 50 volts on the grid of the output
valve. Strictly speaking this will not be the case because the
output valve will start to draw grid current once the voltage
rises above zero and the input impedance will then drop and reduce
the voltage, but you can see the principle.
In fact all the capacitors in the set
needed replacing. Philips use a metal cased block carrying six
decoupling capacitors. Two had been replaced previously. I fitted
four to replace those remaining. I also replaced one of the two
I fitted two modern 10uF 450V
capacitors. A purist would no doubt remove enough of the contents
of the original cases and fit the new parts out of sight.
Next I measured all the resistors. To my surprise all were in excellent condition and very close to the values shown on the circuit diagram.
Philips had made nearly all the
connections with bare tinned wire of something like 20 SWG. After
poking around looking for capacitors etc I found several wires
that needed pushing apart. Once this was done the set was ready
to try out using mains input.
When the set was designed European radio stations were not allocated any particular broadcast frequencies. Most where positioned in the spectrum just as they had been in the early 20s. It was only in the late 20s and during the 30s that concerted efforts were made to divide the broadcast bands into channels and to produce order from disorder. Stations were allocated new frequencies where necessary, so that interference between them was minimised. During daylight hours interference wasn't much of a problem, but night time produced a cacophony of noise as skip changed and broadcast ranges increased from tens of miles to thousands of miles.
In Europe a medium wave channel spacing
of 9 KHz was chosen in order to squeeze in as many broadcasts
as possible. Station broadcast frequencies were then shifted
one way or another to line them up with the new channels.
Once station broadcast frequencies were
agreed set designers produced dials carrying station names. This
Philips set was just a little too early and the dial is calibrated
in wavelengths. 1500 metres corresponds to 200 KHz.
Having determined that the old Philips
could satisfactorily receive long waves I then selected medium
waves. The waveband switch is rather novel in that one pushes
the tuning knob (for medium waves) or pulls it out (for long
waves) to effect a change.
A quick note about replacing a valve...
NEVER just pull off a valve top cap clip or connector, at least
don't do it before you've considered what might happen. Top caps
are usually connected to valve grids and are safe to touch. However,
an early valve top cap is invariably connected to its anode and
this might be between 200 and 300 volts.
Below are pictured the two tetrodes RF valves and the triode audio amplifier
Although I have thousands of old
radio valves it isn't always easy finding the right one, but
after an hour or so I discovered several that were considered
close equivalents to the S4VB with the microphony problem.
During testing I found the aerial worked
fine in the socket marked A2 but not particularly well in socket
A1. Checking the service sheet produced by The Wireless &
Electrical Trader on March 13th 1943 and the circuit diagram
explained why. Socket A1 isn't connected to anything! This socket
was intended to be used by the listener that lived very close
to a large local transmitter and relies on the tiny capacitance
between the A1 and A2 sockets.
AVC was such an important design feature that special valves were introduced. Because changing the bias voltage basically only changed the anode current of a valve and didn't have a dramatic effect on its gain, a method of gain change needed to be introduced. The "variable mu" valve was designed to overcome the problem. This new valve had a different type of grid. Instead of being wound as a coil with the same spacing between turns, the turns were spaced differently along its length. As the anode current changed more or less of the grid became active. Using this principle an amplifying stage could be designed to operate at a given anode current at which the control grid voltage was able to change the amplification. The term "mu" is used to describe the amplification factor of a valve, hence "variable mu" means variable gain.
Because this Philips receiver has
no AVC a strong signal would overload the circuitry and produce
distortion, hence A1 was available if one was listening to one's
local station only a few miles away. Plug the aerial into A2
and tune away from the local station and the set could hear really
weak stations. Because the tuned circuits are of high quality
the set will not then be troubled by the local station. Note
that the word "quality" is an early technical term
referring to the physical construction of the coils. For any
tuned circuit using a coil there will be a resistance, or impedance,
at the frequency to which it is tuned (resonant frequency), and
in fact, a range of impedances at other frequencies lower or
higher than the resonant frequency.
One could wind a coil from thin wire or very thick wire, steel or silver plated copper and the difference between these is described by their quality or "Q".
The lower the DC resistance of the wire the better the Q and the lower the impedance (or resistance at AC) the better the Q. What this means in practice is the better the Q the louder the received signal and the better the rejection of signals that are not tuned in. The Philips Superinductance range of sets, which includes this model, use high Q coils enabling them to operate well even in the vicinity of strong local stations because the narrow passband rejects all but the station to which one is listening. The term "passband" is a technical term which defines the width of a tuned circuit in terms of the voltages at a point off-resonance when the voltage drops to half that measured at resonance.
In theory the Q could be so high that one couldn't hear any modulation on the received station, but in practice especially when several circuits are operated in series as in a radio, the manufacturing tolerances will stagger the centre tuning points. "Stagger" is another technical term which indicates tuned circuits in series are adjusted to slightly different frequencies.
When the Philips set is tuned towards a station one suddenly hears it emerge from the background noise with good clarity. As the station is tuned in it gets louder but also begins to lose it clarity or sharpness and becomes mellow sounding.
I think this explains the quality of sound one associates with old receivers.
Modern receivers are usually designed completely differently. RF tuning as such may not exist because a broadband front end is used which gives a flat gain over the whole range of operation. In these sets the passband is fixed by a special device inserted in the amplifying circuits. These amplifying circuits have a flat-topped frequency response which is wide enough to provide high fidelity reception of a signal so gone is the mellowness of the old designs.
Back to the overhaul of the 834A...
Putting the chassis back into its case revealed a practical problem.
At some time the knobs had been replaced. Not normally a problem,
but the switching from medium to long waves requires one to pull
the tuning knob and vice versa. The replacement knob on the tuning
control had not been pushed in for years because the clearance
to the case wasn't big enough. When it was first fitted it pushed
the spindle in and selected medium waves, but after it had been
pulled out for long waves there wasn't enough clearance to push
it back in again.
I have a theory that most early radio sets that still exist today were put away in the loft when they didn't perform properly because they were still worth lots of money. This set had certainly not been worn out and looks to be in remarkably good condition. Maybe the owner decided to put it away till they could afford to find out why it wouldn't receive the Home Service? After all they'd just paid a small fortune to have a couple of valves and capacitors changed so they certainly weren't going to throw it away; just put it in the loft for a few weeks. That was probably 60 years ago.
As a postscript... The set used a Cossor V6 output valve, but little is listed about this type (I think there is also a "V5"). I wonder if the code resulted in so much confusion that the numbers were changed.