A Mid-1930’s ‘Magic’ Behemoth: Restoration of an RCA Victor Model 15K-1 – Gerry O’Hara Background I recently completed the refurbishment of a Marconi CSR-5 receiver for a friend. Shortly before work on that receiver was completed, he asked if I would be able to restore an RCA Victor 15K-1 receiver as my next project. Quite a different ‘beast’ from the CSR-5, a WWII Canadian communications receiver built for the Canadian Navy, whereas the RCA Victor 15K-1 is a high-end domestic console style set dating from the 1936/37 model year. The cabinet was in poor condition (photo, right), and in need of stripping/re-finishing, but the chassis appeared complete and in reasonable shape from the photos I was sent in advance. To save bringing the large, heavy cabinet over to Victoria from the BC Mainland, and as I don’t have the facility to refinish large cabinets at my house at the moment, it was agreed that I would restore the chassis and the cabinet would be restored by a mutual friend at the SPARC Museum. The RCA ‘K’ series Radios and the ‘Magic Brain’ RCA Victor introduced receivers with a separate RF sub- chassis, marketed as the ‘Magic Brain’, in the mid-1930’s, initially with their models 128, 224, C11-1 and others: “Inside RCA Victor all-wave sets is an uncanny governing unit ... Human in its thinking, we compare it to the human brain. You choose the broadcast - from no matter where in the whole world. Then, watchman-like, it keeps out undesired radio signals. It concentrates on that one and makes it four times stronger. Each tone has higher-fidelity ... in a quality of reception heretofore unequalled" For 1936, RCA Victor upgraded their ‘Magic Brain’ to use three of the newly-introduced metal tube types. The original ‘Magic Brain’, introduced the previous model year, used just two tubes - glass envelope types 6D6 (pentode RF amplifier) and 6A7 (pentagrid converter). The new version used separate (metal envelope) tubes for the Mixer (6L7), Local Oscillator (6J7), and RF
38
Embed
A Mid-1930’s ‘Magic’ Behemoth: Restoration of an RCA ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A Mid-1930’s ‘Magic’ Behemoth: Restoration of an RCA Victor Model 15K-1 –
Gerry O’Hara
Background
I recently completed the refurbishment of a Marconi CSR-5
receiver for a friend. Shortly before work on that receiver was
completed, he asked if I would be able to restore an RCA Victor
15K-1 receiver as my next project. Quite a different ‘beast’
from the CSR-5, a WWII Canadian communications receiver
built for the Canadian Navy, whereas the RCA Victor 15K-1 is a
high-end domestic console style set dating from the 1936/37
model year. The cabinet was in poor condition (photo, right),
and in need of stripping/re-finishing, but the chassis appeared
complete and in reasonable shape from the photos I was sent
in advance. To save bringing the large, heavy cabinet over to
Victoria from the BC Mainland, and as I don’t have the facility
to refinish large cabinets at my house at the moment, it was
agreed that I would restore the chassis and the cabinet would
be restored by a mutual friend at the SPARC Museum.
The RCA ‘K’ series Radios and the ‘Magic Brain’
RCA Victor introduced receivers with a separate RF sub-
chassis, marketed as the ‘Magic Brain’, in the mid-1930’s,
initially with their models 128, 224, C11-1 and others:
“Inside RCA Victor all-wave sets is an uncanny governing unit
... Human in its thinking, we compare it to the human brain.
You choose the broadcast - from no matter where in the whole
world. Then, watchman-like, it keeps out undesired radio
signals. It concentrates on that one and makes it four times stronger. Each tone has higher-fidelity ... in
a quality of reception heretofore unequalled"
For 1936, RCA Victor upgraded their ‘Magic Brain’ to use three of the newly-introduced metal tube types. The original ‘Magic Brain’, introduced the previous model year, used just two tubes - glass envelope types 6D6 (pentode RF amplifier) and 6A7 (pentagrid converter). The new version used separate (metal envelope) tubes for the Mixer (6L7), Local Oscillator (6J7), and RF
amplifier (6K7): “Just as the original Magic Brain set new standards of shortwave reception, so does the new, improved Magic Brain achieve still higher standards...". The use of separate Local Oscillator and Mixer tubes was common practice by many other manufacturers at that time, particularly for their higher-end ‘all-wave’ receivers, where the separated Mixer and Local Oscillator functions resulted in generally improved performance, especially for shortwave coverage. It was this version of the ‘Magic Brain’ that was employed in the 15K-1 model. An article in ‘Radio Age’1, examining what was thought to be a prototype RCA ‘23K’ chassis (never in production), identified that RCA were further developing the ‘Magic Brian’ concept to include automatic frequency control (AFC) on the Broadcast Band. RCA also introduced the ‘Magic Eye’ (type 6E5) for the 1936 season. This
was used on several of their higher-end console models and some table
models. The ‘Magic Eye’ was purported to be an aid to tuning by providing
a visual indication of how well a station was tuned-in and its relative signal
strength. In realty, of course, this was mainly a marketing ‘gimmick’ (as was
the ‘Magic Brain’ title for that matter), as it somehow brought the radio ‘to
life’ by engaging another of the user’s senses – and it ‘moved’ while you
turned the tuning knob – so cool! “It’s alive!...”
Radios with ‘eye tubes’, as ‘Magic Eyes’ are commonly referred to in North America, are still desirable –
providing the ‘eye’ is bright and working well2 - somehow that eerie green glow that changes with the
signal is still ‘magic’ even after 84 years!3
Oh, and not forgetting the (not quite) ‘all-metal’ tubes! – another gimmick? or
a worthwhile technical advance over their glass-envelope counterparts? –
well, RCA’s marketing guys certainly thought so: "Metal Tube radios are
quieter, more sensitive, better toned, superior in every way".
These started
to be
introduced in
1935 by RCA
with great
fanfare over
their
robustness
compared with
their glass-
envelope counterparts, along with the
inherently “almost perfect electrostatic
shield” due to the grounded metal
1 ‘The Should-Have-Been RCA Model 23K: Analysis of a 23-Tube Prototype Receiver’ by Leigh Bassett (Ch 1 and Ch 2) 2 Eye tubes ‘wear out’ relatively quickly – meaning that they become dim with age as the phosphor coating on the target becomes less active over time (usually the tube is still operating but the target is so dim it cannot be seen in normal lighting conditions). Certain types of eye tube are becoming scarce these days, including one of the most commonly-used, the 6U5/6G5, which includes a remote cut-off triode amplifier section (this extended the input voltage range of the tube). The 15K-1 used the less-common 6E5 tube, which had a sharp cut-off triode section 3 Check out this website for different display patterns and more info about eye tubes
pesky external metal shields for RF, IF and small-
signal AF tube applications. After around 85
years, the jury may still be out on that question,
but on balance, they probably were a
technological advance, though perhaps a bit of a
‘dead-end’? Interestingly, the later ‘miniature’
tubes reverted to glass envelopes again – likely
due to ease of manufacture and associated cost
considerations, and it is reported that some
apparently ‘all-metal’ tubes had internal glass
envelopes from the late-1930’s, or even double-
concentric metal envelopes4. A 1942 video of
these tubes being made can be viewed here (Part
1) and here (Part 2). The first metal tubes were
the 5Z4 rectifier, 6A8 pentagrid converter, 6C5
and 6D5 triodes, 6F5 hi-mu triode, 6F6 power
pentode, 6H6
dual diode, 6J7
sharp cut-off
pentode, 6K7
remote cut-off
pentode
(called a ‘super
control’ pentode in the RCA announcement bulletin), and the 6L7
pentagrid mixer. My favourite metal tube is the diminutive 6H6 dual
diode (photo, above) – its just so cute! - and the 15K-1 chassis sports two of them!
The 15K-1
The RCA Victor Model 15K-1 was a ‘top of the range’ console radio marketed in the 1936/37 model year,
which would lighten your wallet to the tune of around $2005. The cabinet styling was maybe a bit more
imposing than the less-expensive RCA 13K, 10K, 8K, 7K and 6K siblings, but somewhat more modest than
the earlier ‘C’ series models. Perhaps only the radio-phonograph combination sets, such as the 15U
model from the same ‘golden years of radio’, were more physically imposing?…
The circuit design also resembles the lower-end models, though with several enhancements (‘bells and
whistles’), mainly aimed at improving the tonal quality and power output of the reproduced audio, eg.
complex tone control arrangements, push-pull 6L6 output stage, and large (12”) loudspeaker, along with
other tweaks, such as more sophisticated eye tube circuitry. The schematic is shown in Appendix 1.
4 A careful dissection of a metal tube (a 6J5 triode) was undertaken by the British National Valve Museum here, where no internal glass tube was found, though glass was used as part of the construction. However, other tube types, or tubes from other manufacturers, could have different construction details 5 Allowing for inflation, this is around $3,700 in 2020 dollars
(6E5), and dual rectifiers (2 x 5Z4) - the latter needed to supply the higher than usual HT current.
The receiver tunes from 150KHz to 60MHz in five bands, comprising:
- ‘Long Wave’ (Band X): 150 – 410KHz;
- ‘Standard Broadcast’ (Band A): 530 – 1,800KHz;
- ‘Medium Wave’ (Band B): 1.8 – 6.4MHz;
- ‘Short Wave’ (Band C): 6.4 – 23MHz; and
- ‘Ultra Short Wave’ (Band D): 23 – 60MHz.
A concentric knob dual-ratio tuning mechanism is
incorporated, providing 20:1 or 100:1 ratios, which provide
good tuning control even on the ‘Ultra Short Wave’ band.
As noted above, the RF circuits comprising RF amplifier,
Local Oscillator and Mixer stages, are all contained on a
detachable sub-chassis, marketed as the ‘Magic Brain’.
Although this is quite a neat design, both mechanically and
electronically, there really is no justification for either the
term ‘Magic’ or ‘Brain’ as it comprises conventional 1930’s
technology and certainly does not ‘think’ – it does not even
have pre-set station selection capability which was the
next ‘big thing’ in domestic set development. Those 1930’s
RCA marketing guys were something else!
Some noteworthy features of the circuit include:
- A separate and completely screened RF sub-chassis integrated with the tuning dial and band-
change mechanisms (diagram above, right, from the Service Manual);
- A claimed novel method of coil switching that puts some coils to double use and leaves two of
the ‘Ultra Short Wave’ coils in circuit at all times, thus simplifying band switching arrangements
and enhancing reliability;
- Separate IF transformer and amplifier/rectifier circuit for the tuning eye;
- A ‘Speech-Music’ switched filter network linked into a multi-tapped volume control to provide
audio bass boost (‘Music’) or treble boost (‘Speech’) that varies with the volume setting, giving a
‘loudness’ control effect, boosting bass frequencies at lower volumes;
- A ‘Fidelity’ control that both changes the tonal characteristics of the audio and the IF
bandwidth: the former via a standard tone control, plus automatic switching out of a tone
correction capacitor and defeating the tone control pot completely on the maximum fidelity
setting, the latter also widening the receivers’ IF bandwidth on the maximum fidelity setting by
switching in a tertiary winding on the first IF transformer; and
Restoring an RCA Model 15K-1 Gerry O’Hara
5
- The cathode bias voltage on the AGC rectifier tube is automatically altered by the band-change
switch, changing the AGC delay when on the LF and Broadcast bands, this reducing the AGC
sensitivity threshold on these bands compared to the Medium Wave, Short Wave and Ultra
Short Wave bands.
Although impressive compared with less-expensive receivers of the day, other features being introduced
by high-end manufacturers of the same period, eg. EH Scott and McMurdo Silver, such as two RF stages
for better sensitivity and image rejection on the higher shortwave bands, volume expansion circuits,
separate bass/treble boost/cut tone controls, and selectable multiple IF bandwidths are lacking. Also,
the phono input implementation on the 15K-1 is rather primitive, necessitating changing two jumpers
on the rear apron of the chassis. Indeed, the chassis on this example had been retro-fitted with a switch
and phono sockets to expedite switching between the radio and phono input.
Restoration Strategy
The answer to one fundamental question needs to be answered prior to starting work on a radio: “What
standard is wanted?” – this can range from simply repairing the chassis to operational condition,
through to a full restoration, comprising a complete strip-down and rebuild such that it looks and works
much as it did when it left the factory. The usual level that I will extend to for a radio is what I term
‘sympathetic’, which is around 75% of fully restoring the chassis. By ‘sympathetic’, I mean that the set
be working close to its original specification, and made to appear generally as it would as it left the
factory, but with some ageing/wear through use, and perhaps retaining a few battle scars acquired
through years of use as intended – both above and below chassis and, perhaps, also the cabinet. This
would generally entail re-stuffing/replicating paper and electrolytic capacitors, and replicating/replacing
out of tolerance resistors where these
components are visible, replacing broken
parts with suitable parts resembling the
originals where possible, cleaning the chassis
and cabinet, repairing broken mechanical
parts/mechanisms and then aligning/testing
the restored set. This level of restoration,
requires significant additional effort to that
needed simply to render the set operational ,
ie. ‘repair-level’ work (replacing only those
parts that have failed to the point that the set
no longer functions properly), or to render the
set operational and more reliable long-term,
ie. including a level of preventative maintenance, such as ‘shotgunning’ (full replacement) of all paper
and electrolytic capacitors and any out of tolerance resistors (but not by re-stuffing or reproducing these
to give an original appearance), in what I would consider ‘refurbishment-level’ work, which may or may
not include some level of cabinet repair/touch-up or refinishing.
In a ‘sympathetic’ restoration, however, I would not expect that every replaced/re-stuffed or
reproduced component, lead dressing and other cosmetic nuances would be indistinguishable from the
original, or that every blemish in the cabinet finish had been removed such that the set looked ‘factory
fresh’ as would be in a full restoration in the strictest sense. Rather, the work would be ‘sympathetic’ to
Restoring an RCA Model 15K-1 Gerry O’Hara
6
the look of the original, ie. the chassis would receive thorough cleaning, the mechanical components
cleaned and lubricated (where appropriate), capacitors would be re-stuffed where possible (the originals
may not be present), reproduction parts, eg. dog-bone resistors, would closely match the general
style/colour code method of the originals, and replaced parts, eg. transformers or chokes, should be
chosen, or adapted/modified, to resemble the style or form of the original, though with some ‘artistic
license’, eg. in the design of reproduction labels if the originals are unavailable or unknown. Also, a
judgement call would be made regarding any modifications made to the set. For example, this chassis
had been modified decades ago to improve the phono-radio switching by fitting phono sockets and
DPST switch on the rear apron so this could be done at the flick or a switch rather than having to use a
screwdriver to change jumpers. This modification could still be useful today, eg. to feed a Bluetooth
receiver or MP3 player into the audio stages. This was, of course, a choice for the set’s owner to make.
Given that this is a high-end radio, the owner decided that a ‘sympathetic’ level of restoration was
appropriate, though decided to retain the phono-radio switching modification.
Preliminary Inspection, Dismantling and Cleaning
The complete chassis, which was liberally covered with decades-old grime and dust (photo, below), was
given an initial clean-up using a paintbrush and vacuum after the tubes had been removed, then the
‘Magic Brain’ RF sub-chassis was removed from the main chassis. I decided to work on the ‘Magic Brain’
first.
The top cover for the ‘Magic Brian’ was missing (one was later sourced for it), likely for many decades,
resulting in the sub-chassis also being in a filthy state – normally this has been kept clean by being
tucked inside the cover.
Restoring an RCA Model 15K-1 Gerry O’Hara
7
The main chassis bottom cover was in place, as was the lower ‘Magic Brain’ cover, and these had largely
mitigated dust infiltration into the underside of the chassis (photo, below).
‘Magic Brain’ Surgery
Cleaning the ‘Magic Brain’ sub-chassis included cleaning all the
old ‘gunk’ (dirty grease and oil) from the tuning and band-
change mechanisms (it was bad! – photo, right) and then re-
lubricating the tuning gang, cleaning the band change switch
wafers with Deoxit/Q-tips, cleaning/lubricating the switch
detent mechanism and various pivot points, and careful cleaned
of the original (damaged) tuning dial and logging scale as a
temporary ‘fix’ until reproduction ones were obtained by the
owner.
The ‘Magic Brain’ has four tubular paper caps in it, the rest are
all plastic-encapsulated lozenge-shaped mica dielectrics
manufactured by RCA. The paper caps on this sub-chassis are
not the same as most of the originals on the main chassis –
600vw rated dark blue/black paper labels over a regular wax-sealed body, of Mallory manufacture (the
main chassis also contained two of these capacitors) – these could be replacements. The four paper
capacitors were replaced, however, as these capacitors will not be seen (as they are totally enclosed by
Restoring an RCA Model 15K-1 Gerry O’Hara
8
the lower ‘Magic Brain’ screening box), they were not re-stuffed. All resistors in the ‘Magic Brain’ unit
were checked and were all within 20% of their nominal values, so were left in place.
The wiring in the ‘Magic Brain’ unit is a
mix of both cloth and rubber insulation
types (photo, left). Two of the colours
(yellow and blue) of the rubber
insulation had perished (crumbled).
The yellow is the AGC line to the RF
amplifier/Mixer stages, and the blue is
the plate of the mixer tube to the 1st IF
transformer (which is at high voltage of
course). Both of these go through a
short umbilical connecting the ‘Magic
Brain’ to the main chassis through an
octal plug/socket.
The cloth-covered wiring passing
through the umbilical was ok, but the
rubber insulated ones had to be
replaced as their insulation had mostly
crumbled away, and the bare wires
were shorting to the chassis where
they passed through a hole on the top
of the ‘Magic Brain’ sub-chassis.
The umbilical is covered in a tight
woven braid that I could not
remove/replaced without destroying it.
Four options existed to remedy this:
1. Replace the rubber insulated wires by running two new cloth wires of a similar colour to
the braid, say brown and green, on the outside of the umbilical, hidden as best as possible
under/behind the umbilical when the chassis is installed in the cabinet;
2. Remove the braid, unsolder all wires from the octal plug, install the two new wires and
replace the braid with black shrink wrap;
3. Remove the braid, install the two new wires and leave as-is.; and
4. Remove the braid, install the two new wires and wrap the wires with self-amalgamating
tape.
Options 2, 3 and 4 would not look original and Option 2 also risked damaging the octal plug, the cover of
which was difficult to remove. I recommended to the owner to proceed with Option 1, as this would
look more original (existing braid left in place and the replacement wires hidden as best as possible)
unless someone looked very closely and spotted the replacement wires on the outside
Restoring an RCA Model 15K-1 Gerry O’Hara
9
(underside/behind) of the braid. I noted to the owner that I would use cloth-covered replacement wires
rather than ‘fake’ rubber-insulated ones6 to blend-in better with the braid.
The next steps to complete refurbishment of the ‘Magic Brain comprised:
- Re-wiring the entire AGC line (all crumbly yellow rubber insulation), above and below the ‘Magic
Brain’ sub-chassis. I used green cloth-covered wire (less obvious
than the yellow original when running alongside the umbilical);
- Re-wiring the plate circuit of the Mixer tube (I used brown
cloth-covered wire – also less obvious than the blue original);
- Reassembled the octal socket on the umbilical (photo, left),
incorporating the new wires, and then chopped-off the old wires
at either end of the umbilical;
- Re-wired the three tube grid cap connectors as the insulation on
all three was degraded and one had a dry joint, and replaced the AGC wiring around the tuning gang;
- Replaced one red rubber insulated wire with red cloth-covered wire (precautionary, as it is an
HT circuit and was starting to crumble at one end); and
- Checked all the remaining rubber
insulated wiring in the ‘Magic Brain’ and,
although showing signs of degradation, I
considered it to be stable enough and
spaced sufficiently away from other
wires, components and the chassis to be
safely left in place.
The ‘Magic Brain’ chassis was then set
aside until it was needed during re-
assembly of the complete chassis.
Photos above/below the cleaned and re-
furbished ‘Magic Brain’ chassis are shown
at the top of page 10.
The owner had decided to replace the
broken and worn original dials (photo,
right) with reproductions to improve the
appearance. The reproduction dials had
been ordered a few days before (from
Radio Daze).
6 By ‘fake’ rubber (insulation), I mean PVC insulated wire, sanded to give a matt finish and then rubbed with some greasy grime – this looks very much like old rubber insulation in good condition
Restoring an RCA Model 15K-1 Gerry O’Hara
10
Main Chassis
The main chassis, less the ‘Magic Brain’ unit, was next on the bench (photo, below of the main chassis
with the ‘Magic Brain’ to its left). Further, more detailed, cleaning was undertaken first, and at the same
time a close visual inspection was made of the components and wiring. Again, some rubber insulated
wiring was present, however, significant degradation was only an issue on the grid cap leads and on a
couple of under-chassis runs. This inspection was followed by the following program of work:
Restoring an RCA Model 15K-1 Gerry O’Hara
11
- Re-wiring of all the IF transformer grid cap
connectors (crumbling/damaged insulation), and cleaning a
huge amount of fluff and dead bugs out of the top of each
IF transformer cap – photo, right;
- On testing, I noted that a section of one of the
Candohm resistors was ‘dodgy’: reading around three
times the nominal value (65 Ohms instead of 25 Ohms),
and reading erratically on the ohmmeter, ie. its resistance
was varying with time. Candohm resistors cannot be easily
repaired, if at all, and replacements are as rare as rocking
horse poop. As it was a low-value resistor I figured its dissipation would be low, so a 1W or 2W part
would likely be ok, and therefore easily hidden below the Candohm unit;
- In addition to a number of replacement capacitors being present, there were several additional
capacitors installed in the chassis compared with the schematic/layout diagram in the service manual.
One had been added as additional decoupling on the ‘Magic Brain’ umbilical connection socket, likely a
‘kludge’ to remedy a symptom of something else causing instability, another as a kludge on the ‘Fidelity’
control, and two more as kludges on the ‘Speech-Music’ control (more on these later);
- The switch on the ‘Fidelity’ control had also been re-wired – I needed to investigate why, but
surmised the switch had failed at sometime and the rewiring was a work-around to render the set
operational;
- Reconstructed the speaker cable/plug using new cloth-covered wire - the old one had been cut
and re-joined twice and the joints taped over with black fabric tape (so ugly!). I also installed a rubber
grommet in the chassis hole where these wires exit to mitigate future
chaffing of these wires;
- Completed the artwork
and printed labels for
reproduction capacitors to
replace the motley selection of
replacements previously
installed in the chassis;
- Made some reproduction
tubular paper and electrolytic capacitor bodies, attached the labels and
coated with amber shellac after stuffing these with new capacitors (see
Appendix 2), and installing in the chassis (photos, above left – the
reproduction capacitor is the farthest one, a re-stuffed original in front, and the photo above right is a
dual low voltage electrolytic under construction):
- Re-stuffed the remaining original capacitors,
and installed these back in the chassis (these were a
mix of 630v and 1600vw types: a re-stuffed 1600vw
one is shown in the photo, right); and
Restoring an RCA Model 15K-1 Gerry O’Hara
12
- Tested several ‘critical’ resistors (in circuit) – most measured within 20% tolerance, but some
higher-value ones were significantly out of tolerance, and were therefore replaced with modern 0.5W or
1W parts ‘disguised’ to blend-in more with a coat of amber shellac.
Someone had added a phono/radio
switch and two phono sockets to the
rear apron of the chassis (photo, right).
I suggested to the owner that it may be
better to just leave as-is as part of the
receivers ‘history’. These types of mods
were very common. Originally, there
were two shorting straps across the two
pairs of screw connections on the
adjacent (original) phenolic connection
strip. Fiddling with these likely became
tiresome for a previous owner and so
had a service technician fit the switch and phono sockets to make it easier to use the receivers excellent
audio stages for his gramophone (and could be used for iPod or iPhone MP3 input these days). The
current owner agreed and so this mod was left in place.
As noted above, the ‘Fidelity’ control had been re-wired and an
additional capacitor added as a ‘kludge’ to render the set
operational without it (yellow arrow, photo, left). This control
would normally perform three functions and is a combo of a
potentiometer and speciality multi-pole switch that acts at full
clockwise rotation of the control (‘Full Fidelity’ position). Thus:
- The potentiometer acts as a regular tone control, with
maximum treble cut at the extreme anticlockwise rotation, and
the potentiometer becomes deliberately open-circuit at the
extreme clockwise rotation, thus defeating the tone control
action completely when in the ‘Full Fidelity’ setting;
- One switch adds in a tertiary winding
on the first IF transformer (circled yellow on
the schematic, right), widening the IF
bandwidth on maximum clockwise travel of
the control to increase the IF bandwidth on
the ‘Full Fidelity’ setting; and
- The other switch cuts out a capacitor
wired across the output tube grids (additional
tone function) in the ‘Full Fidelity’ setting.
These switches were not functional and
someone has disconnected the wiring from
the switch to the IF transformer tertiary
Restoring an RCA Model 15K-1 Gerry O’Hara
13
winding, shorted the windings on the 1st IF transformer to permanently exclude the tertiary winding,
and added an additional capacitor for no apparent reason (I removed it). There is no simple way to
reproduce this rather fancy control, which in its as-found state would act just like a regular tone control.
The options presented to the owner were:
1. Leave as-is and use it simply as a tone control;
2. Fit a NOS potentiometer I had in stock that had a latched push-pull switch fitted. This
would allow the tone control to function (turning the pot) and then pushing the knob
would increase/decrease the IF bandwidth at any position of the tone control; and
3. Try to repair the control by dismantling, inspecting to see what the problem is, and/or
cleaning with Deoxit.
I felt the Option 2 would be a good compromise if the repair to the original control did not work, and the
owner was in agreement with this approach. The NOS pot and switch I had in stock tested ok and I
installed it in the ‘Fidelity’ switch position once the original switch was removed. The switch was wired
to short out the tertiary winding on the 1st IF transformer: when shorted, the receiver would be in
‘Narrow’ selectivity mode and when open it
would be in ‘Broad’ selectivity mode. The tone
control was wired normally and the additional
(switched) tone correction capacitor was
removed from the output tube grid circuit. I
then dismantled the original ‘Fidelity’ control
(photo, right) and found the switch section was
riveted together, though parts of the switch
mechanism could be seen through the hole
where the actuating pin engaged with the
potentiometer. I decided that I would clean
the potentiometer track/slider and then fill the
switch section with Deoxit and leave to soak
overnight sealed in a plastic bag.
Meanwhile I finished the under-chassis recap and remaining resistor testing, replacing ones that exceed
20% tolerance, except for a couple of grid isolation resistors on the AGC line (not critical) that marginally
exceeded this tolerance, and one in an IF can that was also marginally out of tolerance (but also not
critical in value). I also re-wired the phono/radio switch mod as the wiring/soldering was really poor.
Next steps included:
- Removal of the two 10uF parallel-connected can electrolytic (reservoir) capacitors. These had
previously been replaced by a single 20uF ‘Aerovox’ unit, with one of the originals left in place and still
connected(!) – it had been leaking electrolyte around its terminal spigot under the chassis, now forming
a rather crusty mess. A friend had sourced a couple of matching capacitor cans to replace these with - I
planned on using one 22uF 500vw capacitor in one of these cans, and the other would be a dummy unit
to save cutting into it (and preserving the 1930’s electrolytic technology it contained for posterity!);
- Checking out the ‘Compensating Pack’, a potted resistor and capacitor network enclosed in a
sealed metal can mounted near the front of the main chassis. It had been completely disconnected
Restoring an RCA Model 15K-1 Gerry O’Hara
14
(except for the ground wire), and the wires emanating from it taped-up. Whoever disconnected the
wires ‘kludged’ a couple of capacitors and a resistor between the volume pot and the adjacent
‘Music/Speech’ switch to make it work somehow, but it was definitely not doing what it should. The
‘Compensation Pack’ can is spot-welded around the upper edges (above the chassis), and the tags
securing it to the chassis bent around and heavily soldered under the chassis. This would be difficult to
remove to re-stuff the can, and risked damage during this process. In agreement with the owner, I
opted to replicate the circuit outside the can, with the parts hidden from view when the chassis is
viewed from behind (as it would be in the cabinet) by the ‘Compensation Pack’ can itself (see below);
- Replacing the 1Mohm resistor in the tuning eye
flying lead line socket (it was measuring 2.4Mohms
– should be 1Mohm) and re-made the connections
in the socket as they had frayed (photo, left);
- I found a NOS 6E5 eye tube in my stock and
fitted this;
- Cleaned the Music/Speech and ‘Bias’ front
panel switch with Deoxit - the ‘Bias’ switch works in
conjunction with the Band change switch to
increase the AGC threshold (‘delay‘) on the LF and
Broadcast bands. After cleaning, these controls
were both working well;
The overnight ‘soak’ of the Fidelity switch
mechanism worked (much to my surprise!) – both switch sections were now working properly, as was
the potentiometer. Good news indeed. With this control re-assembled, I then:
- Removed the push-switch replacement ‘Fidelity’ control from chassis that I had installed
previously in case the original control could not be repaired;
- Installed the original control in the chassis, along with original mica tone compensation cap, and
re-wired the switches and
potentiometer as per schematic to
provide its original functionality;
- Removed the last two large
18uF can capacitors from the chassis
ready to re-stuff, and cleaned/tidied-up
the associated wiring;
- Found a suitable small tag
board with built-in stand-offs, and
rebuilt the ‘Compensation Pack’
assembly per the schematic onto this
using three capacitors and four
resistors (photo, right);
Restoring an RCA Model 15K-1 Gerry O’Hara
15
- Cut the ‘Compensation Pack’ umbilical where it exited the original “Compensation Pack’ can,
and stripped some of the braid and screen wire back;
- Removed the ‘kludge’ wiring, caps and resistor that had been installed between the volume pot
and the ‘Music-Speech’ switch;
- Installed the tag board on the front of the ‘Compensation Pack’ can with two self-tap screws
through the stand-offs, and wired this up to the volume pot and ‘Music-Speech’ switch as per the
schematic, using original colour coded wiring. Although this tag board will likely not be visible when the
chassis is installed in the cabinet (as it
faces the inside front of the cabinet), I
made a small ‘period’ enclosure (box)
for it, marked with the correct RCA part
number and description and then ‘aged’
it with amber shellac so it blends in
better with the chassis (photo, right).
This box is a neat push-fit over the tag
board, so can be easily removed if
desired, or for future servicing access.
- Before I re-stuffed the can
capacitors, I decided to jury-rig the
(bare) capacitors into the circuit under
the chassis to make sure they had
adequate voltage ratings
(photo, left) – especially
the reservoir cap, as the
highest voltage rating I
had in stock was 500vw7.
- Re-checked my wiring
of the ‘Speech/Music’
(and on/off) switch to
the volume pot, as there
is only five connection
lugs on the volume pot
and the schematic and
layout diagram both
show six. I thought
perhaps the ‘cold’ end of
the pot track was
connected to the body of the pot as a ground (making the sixth connection), but it wasn’t. After a lot of
resistance measurements around these controls, I decided to do a small ‘kludge’ to get around this issue
7 If the voltage across the capacitors on switch-on (surge) was very close to or exceeded this I would have used two 47uF 450vw caps in series plus a couple of voltage equalizing resistors in parallel to give a nominal 23.5uF 900vw capacitor with plenty of surge voltage ‘headroom’
Restoring an RCA Model 15K-1 Gerry O’Hara
16
using a resistor as a temporary fix, as I figured that I would likely have to strip the volume pot down to
investigate further, as it probably would need cleaning anyway – see below;
- Tested all the tubes: the 6E5 was U/S (open filament), as was the 5Z4G rectifier. The other
rectifier was a 5U4G, which, although it has the same pinout as the 5Z4, has a 3A heater and directly
heated cathode, whereas the 5Z4
has a 2A heater with an indirectly
heated cathode. One of the 6K7s
tested as marginal, so did one of
the 6C5s and one of the 6L6Gs.
Another 6K7, although testing good
had an intermittent connection in
the grid cap. Also, one of the 6K7s
had been subbed with a 6J7 (the
sharp-cut-off equivalent of the
remote cut-off 6K7). Apart from
that, the remaining tubes tested ok.
Testing results were as follows (%
emission), together with circuit
position and manufacturer:
6C5 1st audio amp – Philco (65%)
6C5 audio driver – Westinghouse (51%)
6K7 1st RF amp – Marconi (75%)
6K7 1st IF amp – Marconi (62%)
6K7 2nd IF amp – GE (52%)
6J7 eye tube IF amp (should be a 6K7) – Marconi (76%)
6J7 Local Oscillator – Ken Rad (85%)
6L7 Mixer – RCA (59 amp./70% osc.)
6H6 Detector/AGC – RCA (70/76%)
6H6 Eye tube rectifier – Westinghouse (75/65%)
6E5 Eye Tube – Marconi (60% amp. with bright phosphor on target - NOS tube)
6L6 Output – Rogers (65%)
6L6 Output – Marconi (54%)
5Z4 Rectifier – RCA (78/77%)
5Z4 Rectifier – Marconi (72/76%)
Restoring an RCA Model 15K-1 Gerry O’Hara
17
Re-assembly and Preliminary Testing
Before I undertook a full test of the chassis, I first tested the power transformer with all the tubes pulled
to check it was producing correct AC voltages on its secondary windings – it was. I then re-installed the
‘Magic Brain’ onto the main chassis, then tested the output transformer and speaker field coil
resistances – all ok. I noted that there is a small capacitor inside the sealed output transformer
enclosure connected to a small secondary winding - as such it would have low voltage across it, so there
was no need to change it out. I
then undertook some
resistance checks on some
critical circuit nodes (the
Service Manual has a useful
resistance diagram for this
purpose).
Next steps included:
- Installing the tubes,
including replacement (NOS)
6E5 eye tube, and two used
(tested good) 5Z4 rectifiers. I
installed the 6J7 in the non-
critical eye tube amplifier
circuit (it should be a 6K7), the
weak 6C5 as the audio driver,
and the good 6K7 with the
intermittent grid cap in the RF
amplifier stage;
- Fitting a 3A cartridge fuse in the line circuit (in place of the blown 10A one!);
- With the speaker re-attached (photo, above), the chassis was powered up slowly through an
isolation transformer and Variac, monitoring HT voltages…. The chassis ‘came to life’ at around 90vAC
applied to the power transformer primary; and
- Changing one of the #46 dial bulbs that had blown: the other three were ok.
At this stage it actually worked really well – plenty of sensitivity and the audio was really good: powerful
and undistorted. A short video of the preliminary checkout can be viewed here.
Some initial observations:
- The surge voltage on the reservoir capacitor measured around 450vDC, rising to this value
slowly as the 5Z4 rectifiers are indirectly heated. Once the other tubes kicked-in this voltage settles
down to around 385vDC on this capacitor. A 500vw rated part is therefore adequate for this position in
the circuit. The voltage on the two smoothing caps surges to less than 400vDC, and then settles down to
around 260vDC on one and 240vDC on the other, so no problem using 450vw rated parts for these;
- I also tried several NOS 6L7 Mixer tubes to see if there was some issue with the tube fitted into
that stage, but no improvement resulted. Therefore it seemed unlikely this is a tube-related problem
after all.
Further investigation of this issue included:
- Re-cleaning the coils, caps and other parts to
remove any remaining grime, including polishing the
silver-plated ‘coils’ (flat silver-plated strips – follow the
red arrows in photo, right) forming the inductors for
Band ‘D’ on the Local Oscillator, Mixer and RF amplifier
stages;
- Loosening, cleaning and re-tightening various
mechanical ground points;
- Checking for dry or loose joints (none were
found);
- Checking for damaged components: one silver
mica capacitor in the Local Oscillator section had a small
line in the moulding but was not a crack (and that
capacitor is not associated with Band D operation);
- Carefully inspecting the phenolic bandchange
switch wafers (no damage noted), and re-cleaning them
with isopropyl alcohol;
- Carefully inspecting the tube sockets (no
damage noted), and re-cleaning them;
- Re-cleaning the switch contacts with Deoxit on all stages in the ‘Magic Brain’.
On re-testing after the above, no improvement was noted in the 32MHz cut-off frequency, though I
noticed that the oscillator now cut-in again around 55MHz and oscillated happily through to 60MHz, ie.
there was thus a 23MHz window where the oscillator cut out.
Next, I decided to re-check the DC operating conditions. During the soak testing, I had been monitoring
the HT voltage on the reservoir capacitor and this had been holding steady at around 385vDC for the
many hours the set had been working, however, I had not been monitoring other HT supply voltages,
‘downstream’ of the speaker field coil, since the initial switch-on voltage checks. On checking these, I
found:
- These were 20 - 30vDC below what they had read initially;
- This was traced to the speaker field coil: this should have a DC resistance of 1700 Ohms
according to the schematic. When I measured the resistance during the checks prior to
powering the set up for the first time, it actually measured 1910 Ohms. I had put this down to
manufactures tolerance as it was well within 20% of 1700 Ohms, and considered any impact on
Restoring an RCA Model 15K-1 Gerry O’Hara
23
the HT voltage would be acceptable, and although the voltages I measured around the set after
initial switch-on were a bit low, but seemed reasonable; and
- When I re-measured the field coil resistance after the set had been running for a couple
of hours, ie. with the field coil at operating temperature (though barely warm to the touch –
around 32C), it was 2022 Ohms. Before switch on, ie. with the field coil ‘cold’ (20C), it measured
1898 Ohms.
A 200 Ohm higher-than-stock base field coil
resistance, plus a change of some 120 Ohms in
the field coil resistance with a change in
temperature of only a few degrees of the field
coil seemed a bit excessive (my gut feeling –
see discussion below), and suggested to me
that there may be a problem with the field coil
winding, eg. possibly an internal corroded
connection(s) sensitive to temperature.
Removal of the field coil to investigate was not
practical given the way this speaker is
constructed (welded between the frame and
magnet horseshoe – photo right). I was
concerned that if this was the case, the
problem could worsen over time. I tried a
work-around ‘kludge’ of placing a 10Kohm
resistor in parallel with the field coil to render
the combined resistance to around 1700 Ohms
per the schematic. This increased the voltages
downstream of the field coil closer to those I
measured initially and per the voltage chart in
the Service Manual.
With this resistor kludge in place, I checked to see if there were any improvements to oscillator
performance, particularly on Band ‘D’, finding:
- The output of the oscillator was a bit stronger overall (to be expected with more HT volts on the
plate and screen of the tube), though not markedly so;
- The cut-off frequency on Band ‘D’ increased slightly to around 35MHz, but it still died out
completely, and did not re-start until around 55MHz.
I decided to see if this could be improved upon, so I then:
- Re-cleaned the oscillator stage band change switch wafer and contacts (yes, again);
- Using an insulated prodder (a trusty plastic chopstick!), I gently prodded and pried the various
parts of the Local Oscillator circuitry while switched to Band ‘D’ with the chassis running;
Restoring an RCA Model 15K-1 Gerry O’Hara
24
- I noticed some improvement in
the drop-out frequency and oscillator
output by moving the connection to the
tuning gang stator slightly. This is
connected to the band change switch
with a quarter inch hex-headed screw
(follow the red arrow in the photo, right
– here the screw has been loosened).
There looked to be some corrosion
around this screw, so I removed it,
thoroughly cleaned the screw and the
two pieces of metal it joins, applied
Deoxit, cleaned again, inserted the
screw and tightened (plus added some
varnish to prevent movement). I also
adjusted the alignment of this
connection to be as far away from the
hole through the chassis as possible;
- I re-installed the ‘British’ RCA 6J7 tube in the oscillator position (this was the best-performing of
all the ones I had tried previously); and
- Found that the frequency cut-off was now only from 32MHz to 39MHz, with reliable operation
below 32MHz and above 39MHz – so some significant improvement.
However, no matter what else I tried – lead dress, further cleaning, etc, I could not coax the oscillator to
work between 32 MHz and 39MHz – so close (and so frustrating!). I considered changing out some of
the components, eg. mica caps - C12, C24, C26, and C28, plus the three resistors – R4, R5 and R6, though
the resistors all tested within tolerance. C24 is a different type of capacitor than the rest (stubby design,
with large, flat end connections), and is positioned such as to add rigidity to the switch wafer, so
changing this for a modern part would not provide the same rigidity and could cause other problems, so
it was probably not a good idea to change this one out. Also, it would be very difficult to replicate the
‘lozenge’ format of the other mica capacitors as these are generally too small to mill out and re-stuff
with new mica dielectric caps, and would require moulding to replicate properly. Given the minor effect
on the overall performance of the receiver (and on a band that will not be used anyway), my leaning was
towards retaining the original components and sacrificing the possibility of rendering Band ‘D’ fully
working8. I therefore suggested to the owner to consider accepting that the cause is likely age-related,
eg. changes in dielectric properties of the switch wafers and/or a component(s), and leave as-is,
8 My conclusion was that poor physical/electronic design operating at the margins of viability exacerbated with ageing components was the cause of the issue. Apart from the absence of a shorting ring for the unused tuned circuits, the oscillator design itself is far from ideal for these higher frequencies: using pieces of flat metal strip as the ‘coils’, ie. very low inductance, along with a large value variable capacitor (the same as used on all the other bands, ie 11-490pF), results in a very low-Q tuned tank circuit. A better tuned circuit design for frequencies in the range 30MHz -60MHz would use something like a 5 turn air-spaced coil of 16 gauge silver-plated wire tuned with a smaller value variable capacitor, say 10 – 150pF, and the band switch should have a ceramic wafer. This would give a much higher-Q tank circuit, more reliable operation, and more even output across its tuned range
Restoring an RCA Model 15K-1 Gerry O’Hara
25
particularly given that there is nothing to listen to above 30MHz anyway these days9 – the owner agreed
to this approach.
The speaker field coil issue needed more consideration: the set worked well-enough at the slightly lower
HT voltages, and if the field coil resistance did not increase further over time, it should be good long-
term. However, if the field coil resistance increased over time, then the HT voltages could drop to levels
where the set’s performance was significantly degraded, or it could become open circuit, with complete
failure of the set. Options considered here were:
- Leave as-is and see if things remain stable, or if not, deal with it at that time;
- Leave the 10Kohm kludge resistor in place, hidden somehow, to bring the voltages back up to
the nominal values indicated in the Service manual, though again, if the field coil resistance continued to
increase with time, this would only be a temporary solution;
- Connect to the 110vAC tap on the power transformer instead of the 120vAC tap (easily done by
changing the fuse selection). This would increase the secondary voltages on the transformer by around
9%, and, correspondingly, the resulting DC voltages by a similar percentage. This would give a reservoir
capacitor working voltage of around 414vDC (surge of 490vDC), and the heater voltages would be
6.8vAC and 5.45vAC. This may stress the tube heaters, shortening their life a little, and the reservoir
capacitor surge voltage would be close to its 500vDC rating;
- Rewind the field coil (removal of the field coil to do this would be very difficult due to the way
this speaker is constructed);
- Investigate the field coil by removing the protective covering and teasing the connection wires
out – checking the join between the external leads and the magnet wire forming the coil, as this is often
where a problem develops;
- Source a different speaker with good field coil having the correct DC resistance; and
- Replace the electrodynamic (ED) speaker with a permanent magnet (PM) one, replacing the field
coil with a combination of a suitable choke and series resistor.
Considerable discussion then took place with the set’s owner regarding the above, in particular
exploring rewinding the ED speaker field coil, and various options to use a PM speaker, as these seemed
like the most reliable long-term solutions. The latter would, however, negate the care in trying to keep
the chassis appearance original, though if a suitable speaker with a large ‘bell’ over the permanent
magnet could be found, and the output transformer attached to its frame, the choke and resistor hidden
somehow, and the speaker cable/plug retained, it would probably look similar to the original.
9 In the mid-late 1930’s there was the ‘APEX’ stations that could be tuned into on this band – at least in certain large US cities. More on Apex stations here (“….Finally, starting in 1937, several radio manufacturers began to introduce models that could tune all the way up to the Apex bands. The Raco R-S-R Clipper and several McMurdo Silver models were among the first. That same year, RCA introduced its “Magic Brain” series of receivers which had a top band that tuned up to 60MHz. These early radios proved to be insensitive and unstable at those rarified frequencies.”), and here
After much thought and a few calculations (rather than my original ‘gut feel’), I recommended that we
should take careful stock of the situation before taking the rather drastic step of either rewinding the
field coil or converting to a PM speaker. In summary, my thoughts now were that:
- The chassis had now been working for a total of around +50 hours since I first powered it up;
- The DC resistance of the coil when ‘cold’ (room temperature at 20C), as measured now, was
1898 Ohms, so no increase in resistance from when its resistance was first measured at room
temperature before I first powered-up the chassis several days prior; and
- This resistance measurement was less than 12% higher than the 1700 Ohms stated field coil
resistance specification, which would likely be an acceptable manufacturing tolerance.
Given the above, I now surmised that this field coil was probably working more or less as it was when it
was first manufactured, with the increase in resistance once it reached working temperature being
normal. I therefore calculated the expected increase in field coil resistance with temperature: the
temperature coefficient of copper is +0.393%/degree C. Thus for a 15C increase in temperature (35C
seems reasonable for the ‘core’ temperature of the field coil as the surface temperature of the field coil
measured 32C when at operating temperature), the field coil resistance should increase by 15 x 0.00393
x 1900 = 112 Ohms, ie. a resistance of 2012 Ohms when at operating temperature - very close to my
measured 2020 Ohms at operating temperature.
The notes under the ‘Radiotron Socket Voltages’ given in the Service Manual state that the indicated
voltages “…should hold within +/-20% of those shown..” (I measured them with the specified 1000
Ohm/volt meter). Thus, on the Local Oscillator tube, the plate voltage (given as 250vDC) could range
from 200 to 300vDC, and the screen voltage (given as 215vDC), from 172 to 258vDC, which is certainly
the case here - both without the 10Kohm ‘kludge’ resistor across the field coil: 245vDC Plate, 215vDC
Screen, and without: 237vDC Plate, 208vDC Screen.
There was no discernable performance degradation of the set with the slightly lower DC voltages, other
than the Local Oscillator on Band ‘D’ being coaxed into a slightly greater working range at the higher
voltages, which could also be obtained with the bypass resistor fitted across the speaker field coil - a
much less drastic and easily-reversed mod than changing the speaker for a PM one and adding a choke
and resistor. The only reason I had started down the voltage investigation route was because it can be a
factor in oscillators not working as they should, and I was looking for ways the Local Oscillator could be
made to work across its full range on Band ‘D’. Following all the troubleshooting work, I was now
convinced that the cause of the Local Oscillator drop-out window on Band ‘D’, although being influenced
slightly by the HT voltage, ie. this being a little lower than nominal, this was not the primary cause.
The conclusion was that the speaker field coil was actually good after all, and my recommendation to
the owner was therefore to shelve the thought of buying a new PM speaker and replacing the field coil
with a choke and resistor. Instead, I proposed that I keep checking voltages and the field coil resistance
(cold and at operating temperature) during the following days, to see if any changes result over time as
work on the chassis progressed. I expected to clock up another +50 hours or so through a combination
of soak testing, alignment, and general testing/tweaking, observing. If there was no measurable
changes in that time, leave things wired as stock.
Restoring an RCA Model 15K-1 Gerry O’Hara
27
Finishing-up the Chassis Restoration Work
The four can capacitors were installed
and the temporary jury-rigged
electrolytics removed from under the
chassis. The intact (un-re-stuffed)
capacitor can was installed at the end
of the row nearest the side of the
chassis – this will help prevent
damage to the other (re-stuffed) caps
if the chassis is stood on its end as it
will take the load better (I used a
wood chock under the power
transformer shroud to prevent this,
but in the future other folks working
on the chassis may not). After
installing the capacitor cans, I finished
their construction by:
- Adding an extra heat-shrink sleeve over the sleeve
I had fitted previously to the capacitors’ positive
lead when mounting in the can. The positive leads
of the caps are thus effectively triple-insulated by
the original rubber grommet and two heat-shrink
sleeves;
- Soldering small eyelets to the positive capacitor
leads, secured with two layers of heat-shrink. This
looked much more like the original spigot
connections than just soldering to the capacitor
leads (photo, left);
- Epoxying a similar eyelet to the stub of the
original lead on the un-re-stuffed can, again
secured with heat-shrink sleeve so it looked the
same as the other can capacitors; and
- A small length of ‘non-conducting wire’ was
fabricated to connect between the un-re-stuffed
can and the adjacent re-stuffed can (follow yellow
arrow on photo, left). I did this by removing the
conductor wire, clipping it short, cutting in two,
inserting the two pieces of wire into either end of the cloth insulation and soldering to the
capacitor eyelets, so there is a gap in the wire hidden inside the insulation. These two cans
were originally two 10uF capacitors in parallel, forming the reservoir cap: instead I used one
22uF 500vw capacitor in one can.
Restoring an RCA Model 15K-1 Gerry O’Hara
28
I also replaced two of the original cloth-covered wires in the power supply compartment of the chassis
as the insulation on these was suspect, and both carried high voltage: one from the rectifiers to the
reservoir capacitor, and one from a smoothing capacitor to the dual Candohm resistor.
As noted previously, the sets’ owner had ordered a set of NOS RCA tubes, however, he could not find a
NOS RCA 6E5, and was short of both a NOS RCA 5Z4 and 6J7. After some searching through my
inventory, I found a NOS RCA 5Z4 and a NOS RCA 6E5, both ‘JAN’ military issue, and I was sure the
‘British Made’ RCA 6J7 I fitted in the Local Oscillator stage was NOS and military grade also (it came from
the same source…). This made all the tubes now installed in the chassis NOS RCA-manufactured metal
envelope ones as they would have been originally (though the 6E5 is not a metal tube of course!).
Next, I finished re-checking resistor and voltage values after the chassis had been running for around 70
hours, finding:
- All remaining original resistors were still within 20% tolerance except R8. This is the screen
dropper resistor for the 6L7 Mixer tube. I had originally measured it as 65.2Kohms (it should be
82Kohms according to the schematic). The measured value then was marginal at 20%, and it
had drifted a bit lower – surprising, as carbon
composition resistors usually drift higher. This
resulted in a higher than nominal screen voltage on
this tube. Therefore, before I put the base screen
box on the ‘Magic Brain’, I decided to change it out
for a new 82Kohm 1W metal film part (the old one
was a 0.5 Watt dog-bone carbon composition type
– photo, right). Although it is not visible as it is
within the screened box, I gave it a coat of amber
shellac to make it blend in better. The voltage on
the Mixer screen was now correct; and
- The voltage on the screen of the 2nd IF tube measured a little high (142vDC v. 125vDC). I had
replaced the screen grid dropper resistor with the specified 82Kohm part (per both the
schematic and the parts location figure in the Service Manual). My notes showed the original
part measured 150Kohms and was marked as 100Kohms. Increasing this resistor to 100Kohms
would lower the screen voltage to more like the nominal 125vDC specified, however, 142vDC
was well-within the stated allowable 20% tolerance on voltages, so I left its as per the
documentation as there would be
marginal effect on performance (if
any).
I then re-installed the base screen
box of the ‘Magic Brain’ and
placed the chassis back on soak
test. Before I re-fitted the screen
box I removed a couple of broken
metal lugs that were soldered to
the side. I used my mighty Wall-
Lenk solder gun for this (photo, left) – what a tool this is for chassis soldering jobs.
Restoring an RCA Model 15K-1 Gerry O’Hara
29
Alignment and Final Touches
I left the new tubes running in the chassis on soak test for around 10 hours, including several
on/off/cooling cycles before alignment.
IF Alignment
I did not use the Service Manual method for IF alignment as it is difficult to follow and expects specific
1930’s RCA instruments to be used in the set-up. Instead, I:
- Initially set the IF up on ‘Narrow’ using a signal
generator set to 460KHz, peaking all of the IF
transformers three times;
- Peaked the tuning eye IF transformer (per the
Service Manual) for maximum eye deflection;
- Set up the wobbulator with 460KHz centre
frequency; and then:
- Tweaked the first and second IF primary and
secondaries to obtain optimal passband curves on
‘Narrow’ (top photo, right) and ‘Broad’ (lower photo,
right) settings. I noticed that there was IF signal
superimposed on the curve on the ‘scope. The reason
for this is the absence in the circuit of IF bypass caps
on the detector load (R21/R23) – normally a couple
of 100pF – 500pF mica capacitors are connected to
ground here to filter out the IF before the audio
passes to the detector stage. The 15K-1 circuit only
has a single 56pF capacitor in this location -
insufficient to remove the IF signal completely. This
signal was eliminated temporarily while I adjusted
the IF response by connecting a 900pF capacitor
between the diode load (junction between R21 and R22) and ground. Maybe the reason RCA
did not include a larger capacitor in this position was to maximise (audio) frequency response? -
I did notice that the 900pF capacitor did trim some of the highest audio
frequencies slightly.
I checked the IF gain after re-aligning: it was now generating an
additional -3v on the AGC line than it was before these adjustments.
Over -15v AGC could now be obtained on a strong signal using only 6’
of wire in my workshop as an antenna, the set pulling in many stations.
The magic eye was behaving erratically when being adjusted, so I took
its rear panel adjustment pot apart (photo, left), and cleaned the track
and wiper with Deoxit – this fixed the problem and its sensitivity could
now be adjusted properly.
Restoring an RCA Model 15K-1 Gerry O’Hara
30
RF Alignment
The RF stages (‘Magic Brain’) was generally aligned per the Service Manual procedure, but instead of
relying on its method of checking frequencies, I used a Fluke frequency counter for frequency
measurements and a digital receiver to check the local oscillator was not set to the image frequency.
The one thing I don’t like about this chassis is the
‘plunger’ style trimmers on the ‘Magic Brain’. An
RCA 'Type 12636' adjusting tool is specified for
these, which I don’t have. Instead I used a socket
to loosen/tighten the locknuts and a trimming
tool with a right-angled stiff wire end to pull/push
the plungers (photo, right). This worked ok, but
not ideal (a bit fiddly), especially for the Mixer
stage trimmers, some of which are very close to
the vanes of the tuning gang when un-meshed -
which is where they are when adjusting the
trimmers of course, ie. at the top end of each
band.
After alignment I noted that Band ‘D’ (Ultra Short
Wave) was now working slightly better than
previously – the local oscillator now only dropping
out between 38.5MHz and 43.5MHz, ie. over an
approximate 5MHz window. I am not sure why
this is different than before – maybe because the
bottom cover was now fitted to the ‘Magic Brain’
or due to the alignment adjustments?
The scale accuracy and sensitivity is excellent on all bands: the only bands that needed much adjustment
were Bands ‘X’ (Long wave), ‘A’ (Broadcast), and Band ‘D’ – the others were almost spot-on, both on
frequency and RF/Mixer stage alignment.
The frequency stability is excellent once the set has been running for 30 mins or so, even on Band D.
There is a slight drift during the first few minutes after switch on, which is normal. Fitting the top cover
on the ‘Magic Brain’ will likely improve this, as will operating the chassis horizontally - the chassis needs
to be on its end to make the IF adjustments, meaning heat from the rectifier tubes warms up the coil
cans and other parts of the ‘Magic Brain’ sub-chassis more than it normally would. I had the chassis set
horizontally on the bench for the RF adjustments so this heating effect would not affect the ‘Magic
Brain’ alignment.
Following re-alignment, the chassis was left on soak test for several more hours. HT voltages were still
holding steady, within 3% of nominal as indicated in the Service Manual. The total chassis on-time was
now approaching 80 hours since first switch on. All controls were working properly and temperatures of
transformers and field coils were all nominal. With this, the base plate was re-installed on the main
chassis.
Restoring an RCA Model 15K-1 Gerry O’Hara
31
Finally, I touched-up some small rust spots on
the speaker magnet frame using a silver
'Sharpie' pen (photo, right), cleaned the knobs
and re-fitted these to the chassis. I also tried
playing an MP3 file using my iPhone through the
phono input and it worked well.
A short video of the chassis working after
alignment can be viewed here:
Closure
The RCA 15K-1 is a great-performing receiver –
the sound quality is excellent and, for a
domestic receiver, it has plenty of sensitivity
and selectivity10. Image rejection is, inevitably,
not the greatest on the higher frequencies,
especially on the ‘Ultra Short Wave’ band,
though, for what the set would be used for on
these frequencies, this was probably not an
issue. With the addition of a BFO it would have
made quite an acceptable communications receiver.
At the time of writing this article, the cabinet was still being worked-on in Coquitlam. Also, the
reproduction dials were still on order, and these will be fitted by a friend who is working on the cabinet
(his mechanical skills are much better than mine anyway!). I will add an addendum to this article
covering the cabinet restoration at a later date. I hope one day to see and hear the complete radio
working!
10 This is the second RCA 15K-1 I have restored. The other was for a SPARC Museum customer in 2013. That also sounded excellent. I have also restored a couple of RCA 10K-1 receivers – my recollection is that they also were good performers, though having a single-ended 6L6 output stage, these did not pack the same ‘punch’ as the 15K-1