R109A Receiver Commissioning
I decided to get this
old receiver working, so carried out
some cursory checks before applying power and noticed that each
valve filament has a limiting resistor of 71 ohms connecting
to the 6 volt supply except for the first RF stage (either
V1b or V3a depending on the variant... mine is very clear because
it has V3a written on the chassis). The valve in place is actually
an ARP12 so is wrong, but I checked its filament and it was OK,
making me believe that the receiver has never been powered up
in this state. Fortunately, I have lots of SP61 valves so I plugged
in one of these at V3a. I also checked my stock of vibrators.
I have two 6 volt 4-pin types so I fitted one of these.
My usual plan when tackling
a new project is to find out whether anything really nasty has
happened in the past that makes refurbishment too difficult and,
being fairly confident after a good look around the chassis of
not wrecking anything, I applied 6 volts from a current-limited
power supply. The first problem was a lazy on/off switch but
repeated waggling got it working. With any valve equipment the
heater/filament cold resistance is pretty low so the voltage
dropped to virtually zero at switch-on but, by gradually increasing
the current limit, the voltage started to rise. At a consumption
of 2 Amps the current was oscillating somewhat but the vibrator
(to my surprise) had started up and after plugging in headphones
and waiting a few moments I heard a whine from unsmoothed vibrator
I found all the switches were
lazy but responded to waggling. I connected a long wire and after
some further switchery I heard a strong SSB signal around 4MHz.
The BFO turned on after some more switch waggling and resolved
the sideband signal. Turning the wavechange switch was difficult
but it did operate after some persuasion and after waggling I
heard lots of strong AM broadcast stations.
Block diagram of the
What about that 2 Amps
of current being consumed? At first, for a battery operated receiver
2 Amps seemed high, however the SP61 consumes 600mA and the other
valves 50mA each so that accounts for 950mA. If the HT is 150
volts and let's say 20mA is consumed, this equates to 3 watts
and at 6 volts would account for another 500mA, but of course
inefficiencies in the vibrator HT supply and the front panel
lamp (I checked and the bulb was marked 0.05A) might add enough
for a total current draw of 2 Amps.
The next step is to familiarise
with the various components and to identify any that need replacing.
In doing this I spotted a small capacitor buried in the chassis
marked "RS 0.047uF". My guess is that this might be
a replacement for C11b (it was), the condenser used for coupling
audio to the control grid of V2b, the output valve, even though
the original is supposed to be 0.002uF. Another condenser that's
been replaced (or in fact parallelled) is either C19a or C19b
located in the DC output from the vibrator PSU (in fact C19b).
Again the choice is far from the original rating, being 100uF
x 250v. I measured this and found it was 68uF but with an ESR
of >20 ohms (Note: My ESR meter cannot read above 20 ohms
so it could be virtually anything but it should be around half
an ohm). I also checked a few other electrolytics. All were either
open circuit or a few ohms with no capacitance measurable. None
will not stop the receiver working but will result in reduced
overall gain and noise on the LT and HT rails.
Unless they fail in a current
avalanche mode the majority of the wax-paper condensers may be
good enough to leave in place. The exceptions are those used
in high impedance areas of the circuit (eg C11b which has already
been swapped). No AVC is used in the R109A but there may be the
odd condenser associated with the RF gain circuit that will need
As documentation for the R109
is limited, and for the "A" version, misleading in
that the picture purporting to be an "A" is actually
that of the "C" version!, I've marked up the pictures
below to identify components.
Apologies for the poor circuit
diagrams. I'll look out for something better.
The next step was to replace
the two HT condensers (or three if you include the 1970s addition).
I decided on a pair of 6uF sealed metal-cased capacitors because
these had a very low ESR and not electrolytics. I removed the
two clips from the chassis because both old condensers are grounded,
not to chassis, but to HT minus, which I measured to be about
13 volts. Connecting 6 volt power brought the receiver to life
with whine-free audio. Tuning to 80m I heard a decent AM signal,
but on 4.0MHz instead of 3.62MHz and strong SSB at higher points
on the dial. Clearly the range is skewed. I thought at first
it was to move top band further from the tuning limit, but obviously
that's wrong because to do that would need the range to be tuned
downwards, not upwards. Before tackling any discrepancy in tuning
there are a couple of important steps.. firstly the dial mechansm
needs to be checked in case it's been misaligned with respect
to the main tuning condenser, and secondly I'll need to confirm
the whether the IF hasn't drifted or become mis-tuned from its
correct frequency of 465KHz. The IF transformers each have two
adjustments which are dust cores fitted with brass screws. Initially,
an audio power meter can be used and all six dust cores adjusted
for maximum output. Once this has been done the BFO can be zeroed
to 465KHz. Optionally the BFO could be offset because both 40m
and 80m bands use Lower Sideband.
There are several methods of aligning
an IF amplifier, really depending on the test equipment you have.
Years ago a wobbulator using a CRT may have been pressed into
service. Another method was to meticulously and individually
adjust each IF amplifier stage according to detailed instructions
to give the desired overall response.
Any receiver using a crystal filter
is a special case because the crystal resonance characteristics
would dictate the final IF which may be up to a kilocycle away
With a simple receiver like the R109
the easiest way to align the IF strip is to inject a 465KHz AM
signal into the aerial connector and, with an audio wattmeter
or an old AVO set to AC volts connected to the headphone socket,
tweak all the IF coils for maximum audio output. This is my preferred
method, however I like to complete the alignment using my DSA815TG
spectrum analyser which will show up any odd effects.
This "A" version of the R109
has no AVC so careful manipulation of the signal generator output
and the volume control is necessary during the initial set-up.
In fact, although it's called a volume control on the panel,
the R109A control is actually an RF gain control.
You can see opposite that all the IFT
primary coils (L7) are identical as are all the secondary coils
The BFO (L9a) is adjusted by zero-beating
it against a 465KHz CW output from a signal generator.
RF alignment looks to be straightforward
once I've identified the trimmers and dust cores. The R109 and
the A version differ slightly in the labelling of the tuning
condenser sections. Much like the WS19, trimmers for the higher
range, once adjusted are not touched because these are used in
parallel with trimmers for the LF range. Sufficient adjustments
are provided by trimmers and dust cores for aligning both ranges
to conform to the start and end of the dial markings.
IF alignment was not easy. I turned
on my Wavetek digital signal generator and set it to 465KHz AM
around 10mV and connected it to the R109 80 ohm aerial connector
and ground expecting to hear at least a weak off-tune signal..
but nothing, so I increased the output and only when this was
1000mV did I hear anything. This was a rough signal that didn't
vary as the Wavetek tuned up or down, so I turned on my old Marconi
analogue TF2008 and tried this. Unlike the digital Wavetek, the
TF2008 tuning knob can be rotated backwards and forwards around
the expected frequency to determine if a signal can be heard.
Success... I heard a single sharp tuning point in the right vicinity
so turned on a frequency counter and checked the TF2008 output.
It read 453KHz which was so low that my button pushing of the
Wavetek to around plus/minus 5KHz hadn't been enough. Maybe,
if I'd used FM it would have worked?
Having discovered the R109 was nicely
aligned to 453KHz (there was a single sharp response) I plugged
in the Wavetek, tuned it to 465KHz and set the output to 100mV
AM then tuned the two cores in IFT1 which peaked a weak but tuneable
note. Moving to IFT2.. this also tuned and I could reduce the
input somewhat. Finally IFT3 cores both unscrewed to tune in
the 465KHz signal. By now the RF input level was sensible and
the audio wattmeter plugged into the headphone jack socket was
peaking nicely. After another couple of rounds I was happy that
the IF strip was OK. Next, the BFO which had worked previously
to resolve SSB. This must have also been set to around 453KHz
and sure enough, switching off modulation and unscrewing the
core of L9a a very long way allowed me to zero-beat the BFO to
What must have happened in the past?
Without an accurate signal generator one can tune to a strong
signal, then tweak each IF core in turn to increase the audio
level. Then retune the receiver slightly to again add some volume,
then return to the IF strip and continue tweaking. I imagine
that the gain of the IF strip will increase slightly as it's
tuned lower in frequency so repeated twiddling will increase
the recovered audio, but of course you'll end up miles (several
KHz at the IF is miles) away from 465KHz. The final tweak must
have been retuning the BFO.. easy enough.. just screw in the
core and it's now compatible with the wrong frequency.. 453KHz.
Does it really matter what the final IF response is? In the case
of the R109 it's not greatly important except the dial readings
will be out somewhat, and if you tweak the local oscillator to
fix this then continue to tweak things you might be quite happy
if your interest is the 40m and 80m amateur bands. Aligning the
receiver to these bands would be easy but over the whole of the
two ranges you would probably end up with good reception at one
end and poor reception at the other end.
For the R109 to work correctly, as the
designer intended, providing a uniform response across each of
its two ranges, 465KHz must be the IF. This is because many of
the components in the front end were selected to conform precisely
to the choice of 465KHz, in particular the tuning condenser and
the oscillator padder condensers. Of course component tolerances
and ageing will somewhat degrade the final results but correct
alignment will always be the best option.
Now for the front end alignment...
there's some vagueness about this because the trimmers and coils
for R109 and it's "A" variant may have different locations.
The first step was to free up the wavechange switch and to clean
its contacts. A few squirts of switch cleaner worked quite well
and good enough to proceed with alignment. Clearly, although
I can receive signals, it's very noticeable that the background
noise across each of the two ranges varies considerably indicating
poor alignment. At every point on the tuning dial it's essential
that the oscillator, mixer and RF amplifier coils are all precisely
tuned. This is made possible by alternately tweaking the coil
cores at the LF end of each band and the trimmers at the HF ends.
By doing this several times you eventually find a constant background
noise when tuning across the dial, and more importantly.. dial
readings will be correct.
The LF range was miles out and it became
obvious that the various cores and trimmers had been haphazardly
twiddled during the life of the receiver in much the same fashion
as the IF adjustments. Little by little I managed to line up
each range (not easy because the positions of the coils and trimmers
for the two ranges on this "A" model are switched around).
Some adjusters were so far out I had to connect the signal generator
directly to the top cap of the RF amplifier.. bypassing its tuning
coils. The LF range tuned something like 2.5 to 4.5MHz instead
of around 1.8 to 5MHz so that the receiver wouldn't initially
tune either my 2MHz or my 5MHz test signals. The HF range oscillator
wasn't quite so bad, but couldn't at first tune my 12MHz test
signal so I had to initially use 11.5MHz to get the oscillator
lined up to the dial markings. I've marked up the locations of
the coils and trimmers, each with the relevant dial setting for
adjustment in this picture.
Note: Four coils and one trimmer are hidden away.
I finally finished the front
end alignment. Very easy once I'd identified the correct coils
and trimmers. The hardest part of the job was seeing the dial
which isn't illuminated. There are three or four electrolytic
condensers that measure zero capacity. The only problem I can
find is perhaps not enough reverse bias so that minimum volume
isn't low enough. In this R109A, designed primarily for CW reception,
the volume control was omitted and a new RF gain control fitted
in its place (although still labelled as volume control). Not
only is minimum voume too high, but when listening to SSB you
can't reduce the RF sufficiently to clearly read it. My guess
is a bad condenser.
Above, the PSU parts responsible
for smoothing and establishing the correct HT, bias and LT for
the R109A. The condition of the wax paper condensers is typical
of what's to be found in WW2 equipment. The reason for this investigation
was to sort out the rather low value of the bias supply. Access
to these components is extremely awkward unless fairly major
dismantling is undertaken, however I took a short cut by unscrewing
the upper rectifier and the bias rectifier. Then I unscrewed
the two 6BA screws securing the upper terminal strip (self-tapped
but locked in position as are the rectifier nuts). Then it was
possible to bend the end plates sufficiently to withdraw the
upper HT rectifier to access R4j, C12h, C12j, C20a and C20b.
After fitting four new capacitors
and a pair of new 270Kohm resistors (R4h + R4j)I tested the receiver,
expecting to find the audio control reduced the output but, alas,
the set was even more sensitive and the minimum audio level was
much louder. I heard very strong amateur sideband on the 60m
band, but so strong was the signal, even with minimum RF gain
I couldn't resolve it cleanly. Looking at the circuit diagram,
the problem must be C4m. The bias voltage marked VB1 below reads
minus 15 volts with the RF gain at minimum and minus 17 volts
at max. C4m (at least) must be the leaky suspect?
This is a simplified circuit
for the manual gain control arrangements. I've not included the
HT components which share the same transformer shown in the previous
drawing of the power supply.
I removed and tested C4m,
which was indeed a bit peculiar. Over the years I've found that
some old condensers not only leak when HT is applied but some
seem to cycle between leak and no leak (or little leak). I guess
this is because of a virtual almost-short-circuit within their
innards. C4m was in this category so I left it out of the circuit
and made measurements without it in place. As I'm unsure of the
working conditions of the J25 rectifier I swapped this for a
standard 1N5402 silicon diode. This gave me a potential at its
anode of minus 70 volts (VB w.r.t. chassis). This feeds R1k,
a 1Mohm resistor (actually marked 910Kohm) and then R10a, a 1Mohm
potentiometer to chassis whose wiper feeds the control grids,
via R1e, of V1e, the 1st IF amplifier and V1f, the second IF
amplifier each via a 1Mohm resistor (R1c and R1f actually marked
820Kohm). Remember that in these bias circuits, virtually zero
current should flow through the control grid circuit unless via
a leak to ground.
The 70 volt bias supply VB will
have a maximum grid bias level of 35 volts (VB1), being half
the total due to the two equal 1Mohm resistors forming the load
(R1k + R10a) with the wiper at the live end of the pot. Let's
assume that 70 volts is the correct value.. but 35 volts appears
to be inadequate to cut off V1e & V1f.. so what's wrong,
because looking up the spec of the ARP12 I see this should cut
off when its grid is minus 20 volts so minus 35 will be plenty.
Following the circuit from R10a this goes to the two valves already
mentioned.. V1e and V1f whose control grids are each buffered
by 1Mohm (R1c & R1f) actually 820Kohm resistors) and decoupled,
by 0.1uF condensers (C4f & C4j..no doubt leaky). The control
grid of the Mixer (V1d) is coupled to the anode of the RF amplifier
through C7g, a 150pF condenser. Could that also be leaky? No,
I checked V1d top cap and it was sitting at minus 1.5 volts.
Now comes a practical problem..
both V1e and V1f control grid decouplers are inside the IF screening
cans so will be difficult to swap if they're leaky. Fortunately,
some guesswork can be taken out of the investigation because
I can simply measure the voltages at V1e and V1f top caps. If
these track the RF gain control wiper then the decoupling condensers
are OK. Well.. the top caps read minus 10 and minus 11 for V1e
and V1f which sadly means bad condensers inside the 1st and 2nd
IF cans. I say "sadly" because the IF cans are soldered
to bases screwed to the chassis and to get at their innards means
a lot of dismantling, particularly to get at the 1st IF can in
the centre of the chassis. Is there any option, bearing in mind
I only need to turn down the audio volume (or reducing IF gain
in order to resolve SSB)? Possibly there is a simple solution...
I initially said that the bias voltage was minus 35 volts due
to the load imposed by R1k and R10a, so what if I reduced R1k
from 1Mohm to say 220Kohm? The bias voltage would move from minus
35 to minus 57 volts. Each of the two top caps is at minus 10
volts and so the loss in the respective bias resistor is about
20 volts so we need to double the bias supply at the volume control
pot from 30 volts to 60 volts to bring each top cap to minus
20 volts (assuming this is the cut-off voltage). In this state
the loss in each bias resistor will be 40 volts and the bias
will be minus 20 volts. This means that bridging R1k to bring
it down to say 200Kohm will do the job of reducing the audio
to a low enough level. Adding 220Kohm in parallel with R1k should
A second solution also looks
possible. Bearing in mind that the available HT from the same
transformer winding feeding the J25 rectifier is around 150 volts,
then this level of voltage might be available as a negative bias
voltage? Looking at the circuit diagram, you'll see that the
J25 is fed via a pair of condensers, C12c & C12h which isolate
any DC connection to the HT circuit. In fact the bias voltage
VB is referenced to HT- or chassis via the two resistors R4h
& R4j. These are equal in value, so the reference point at
their junction will be about half the theoretical maximum available
from the transformer... or as I measured previously 70 volts.
In fact using the J25 rather than a modern 1N5402 this works
out at nearer 60 volts. What then, if R4j was smaller than R4h..
say if I shunted R4j with 1Mohm? I think the bias voltage VB1
might be close to 40 volts using the J25 rectifier. The figures
are a bit woolly because we're dealing with indeterminate leakage
currents in the bad condensers C4f & C4j, however with a
little experimentation swapping or shunting R4h & R4j the
volume control (or IF gain control) will enable SSB to be resolved.
In summary, using J25, which
appears to work satisfactorily, the measured loss from it compared
with a 1N5402 is about 10 volts so, in the first method for fixing
the problem of too much overall gain, I might need to use in
place of R1k, not 220Kohm, but maybe 180Kohm? The downside of
this is that I might need to decouple the bias supply a little
more to compensate for the lesser smoothing effect of the lower
value resistor. Time to try the two solutions...
I found that bridging R4j with
180Kohm and R1k with 330Kohm enabled VB1 to establish a range
of minus 56 to minus 63 volts which brought the minimum RF gain
down to an acceptable level. With these additions I measured
the DC voltages, VB4 & VB5 at V1e and V1f to be minus 14
and minus 11 volts respectively, which seems to be sufficient
negative bias to protect ones ears. Incidentally, on that subject,
the crash limiter which uses another metal rectifier, works a