Emerson 838 hybrid

valve/transistor radio

The Emerson 838 is a transitional design in more ways than one.

It came at the end of the valve era, as transistors were starting to
become widely available and thus uses both. Many of its components
are mounted on a riveted phenolic board but it also has a metal
chassis, representing the fact that it was introduced just before sets
began to be built using printed circuit boards.
The Emerson 838, with its punched this set is not as compact as the Re- ling, there's little externally to distin-
and riveted phenolic board chassis and gency TR-1, shown next to it for com- guish them. The 838 comes in several
metal frame, sits between the older all- parison, it's impressively small for a different colour combinations.
metal chassis designs and upcoming hybrid set. I have the silver set shown here,
printed circuit models. We covered the all-transistor Regen- which is also available with a red back
All the RF stages, the detector and cy TR-1 set in our April 2013 issue; see and tuning knob, one in a maroon case
the audio preamp stage are valve-based siliconchip.com.auhkrticle/3761 The with a gold faceplate and one in cream.
while the push-pull Class-B output two sets were contemporaries, with You can see photos of other versions
stage is based on a pair of PNP tran- the TR-1 (the first all-transistor set) of this set at www.radiomuseum.org
sistors. Despite the use of transistors, released in late 1954 and the Emerson
the loudspeaker is still transformer- 838, in 1955. Construction method
coupled. The Emerson 838 was an evolution Major components such as the IF
While the use of valves means that of the all-valve 747. Besides the label- transformers are mounted using twisted
98 SILICON CHIP Australia's electronics magazine siliconchip.com.au
 41b. it fit
4 6 * 4 6 4r 6

44446
694.4 4 4 4 4 * 4444
4444444486**
• 6.: * it * 4 4 4 * 6,i4.
468444444444,
4***884*4444444
4-66.4*44444 4 *844
44444984448* 94.4
44444.89k.vr44 44444
44.4444 6 6 * 444
.** a ti 44
4444 4. ti"

The Emerson 838 (153 x 90 x 33mm) shown at left with the Regency TR-1 (76 x 127 x 32mm) to its right. Considering the
Emerson 838 used three sub-miniature valves, compared to the all transistor TR-1, its size is quite impressive.
metal lugs and the valves insert into in- were common by 1955, with the only fell out of favour in larger sets after
line valve sockets specially designed to real difficulty being in how to obtain the 1940s; while subminiature battery
contact the thin wire connections of the an appropriate voltage to power the pentagrids (1E8) and triode-hexodes
miniature battery valves. output stage. (2G21) were available, their conver-
Likewise, the two transistors insert The solution was to use a 4V "A" sion conductances are significantly
into chassis-mounted sockets. Most battery rather than the more typical inferior to that of the 1V6.
minor components are wired p oint-to- 1.5V type and compensate by con- Also, the 1V6 has only about half the
point, either to socket/IF transformer necting the three valve filaments in conversion gain of its 1R5 B7G cousin.
contacts or to chassis eyelets. series, so they could also run from Given the 1V6's superior performance
Like some other sets of the era, many this 4V supply. to its subminiature alternatives, it's
minor components are fitted to a "Cou- Dispensing with the output pentode no surprise that the 1V6 dominated
plate"/ "Printed Electronic Circuit" also removed the need for its biasing commercial battery valve designs of
(PEC), an early method of packaging circuit, so there's no wasteful back this era.
components onto an encapsulated bias resistor, as there was in the 747. While pentagrids and triode-hex-
substrate. As it's buried behind other The set uses a ferrite rod antenna, odes rely on the oscillator's signal di-
circuitry, you can't really see it in the moulded into the top of the case. The rectly modulating the electron stream
photos. tuned antenna circuit feeds the signal from cathode to anode, the 1V6 relies
These can crack over time, or be- to the mixer section of the converter, a on the coupling between the two sec-
come damaged but replacements for 1V6 triode-pentode. Triode-pentodes tions for LO (local oscillator) injection.
the more common PEC assemblies are
available online. If you can't find a re-
placement, in the worst case, it is pos-
sible to make a substitute using more
modern assembly techniques.
The "A" battery fits into a conven-
tional spring-loaded bay retained by a
slide cover while the "B" battery (also
behind a slide cover) uses a snap fas-
tener identical to those on the familiar
PP9 transistor radio battery.
Circuit description
Rather than the conventional 1AG4
output pentode of its Model 747 pre-
decessor, the 838 uses a push-pull
transistor output stage.
This significantly improves battery
life as it eliminates the 1AG4's con-
stant 40mA filament current and 3mA
HT current. The "A" battery operating
current falls by 25% but the "B" battery Inside the Emerson 838 case everything is packed neatly. The antenna in the set
current drops by over 50%. is directional, so you might be able to get better reception over its 540-1620kHz
Transistor audio amplifier designs range by rotating the case.
siliconchip.com.au Australia's electronics magazine OCTOBER 2018 99
 Aside from the use of the triode-pentode, it's a conven-
tional converter stage. The tuned signal is fed directly
to the converter's signal grid. Bias for this stage, derived
t:o
from the AGC circuit, is series-fed through the antenna
winding.
The oscillator is a little unusual; the expected capacitive
coupling from the top of the oscillator's tuned winding is
absent. Instead, an open-ended coil winding is used, using
parasitic capacitive coupling between the grids.
Grid resistor R3 (at 1MQ) is much higher than usual,
reflecting the generally lower voltages and currents in
subminiature valve circuits.
The triode's anode current is supplied via the oscilla-
tor coil's primary and the mixer's anode via the tuned pri-
mary of first IF transformer T1. Its secondary, also tuned,
feeds the signal to V2, a conventional sharp-cutoff pentode
(1AH4). Despite its small size, it gives more gain than the
larger B7G 1T4 work-alike with a 45V supply.
The IF amplifier does not receive gain control from the
AGC circuit. That's a result of the set's series filament
connection.
Since each filament is some 1.25V more above ground
than the previous one, series-connected filament designs
demand some tricky AGC action. There's an excellent
description of this on pages 1114-1115 of the Radiotron
Designer's Handbook.
Emersons' designers have picked the elegant solution
of "contact potential" bias with no external gain control.
Grid resistor R4 (10MQ) allows V2's grid to drift weakly
negative and provide self-bias.
I thought that this might also allow grid rectification on
strong signals and thus provide its own local AGC but in
later testing, I was not able to find any evidence of this.
Unusually, the second IF stage is neutralised by 5pF
capacitor C12's feedback from the valve's anode to the
"cold" end of the first IF transformer's secondary. This is
odd because pentodes generally exhibit very low anode-
grid capacitances and do not usually need such a high
neutralisation capacitance.
The 1AH4's Cg_a is just 0.01 pF but note that C12 forms
0 a capacitive voltage divider with 2nF bypass capacitor
C11, reducing its effectiveness, hence the relatively high
value. Note also 22nF capacitor C3 from the bottom end
of the antenna to ground, which is necessary to cancel
out feedback in the overall circuit wiring in this tightly-
packed little set.
V2 feeds its amplified IF signal to the tuned primary
of second IF transformer T2 and T2's secondary delivers
the IF signal to the diode section of V3, the demodulator.
The AGC signal is derived from the DC component of
the demodulated signal, fed back to the grid of converter
V1 via the resistive divider formed by R1/R2.
The AC component of the signal is filtered out by C3
(it's also an RF bypass capacitor, as mentioned above).
Since the "cold" end of the second IF transformer is re-
turned (via R6 and R5) to the valve's filament, there's no
delayed AGC effect.
The audio signal at the wiper of volume control pot R5
is AC-coupled via C16 to the grid of V3's pentode section.
It gets bias from the negative filament terminal of V2,
around -1.2V, via 5.6MQ resistor R7. V3, a 1AJ5, is basi-
cally a subminiature version of the B7G 1S5, with about
80% of the gain for a 45V supply.

100 SILICON CHIP Australia's electronics magazine siliconchip.com.au

 Tuning gang

Output

Above: labelled bottom view of the 838 chassis showing the two IF transformers, output transformer and tuning gang.
Below: labelled top view of the chassis. The large 50pF ceramic capacitor (C21) just under the volume control bypasses
the 4V LT supply, while the smaller 8pF ceramic next to it (C19) bypasses the 45V HT supply.

Audio output stage

Audio preamplifier stages ideally have anode load and primary and low-impedance, tapped secondary.
screen dropping resistors in the megohm range. These very The circuit shows Q1 and Q2 as proprietary Part No.
high values hit the "sweet spot" between increasing gain 815003. This set's devices were 2N34s, a grown-junction
(with increasing load resistance) and decreasing mutual germanium PNP audio transistor type.
conductance (with lower anode/screen currents). Crossover distortion is minimised by the biasing net-
But this valve needs to deliver sufficient current to drive work of resistors R10/R9, providing the usual 150mV of
the following Class-B transistor output stage. The screen forward bias to both bases. Unlike later designs, there is
voltage of 30V gives V3 a mutual conductance of about no shared emitter resistor to improve bias stabilisation
300uS (microsiemens), enough to provide both useful volt- and add local feedback.
age gain and an adequate current. The transistor collectors feed push-pull output trans-
Transformer T3 matches V3's high anode impedance to former T4, with C20 providing a top cut function. T4's
the low input impedance of Q1/Q2, with a high impedance secondary feeds the im speaker directly as there is no

The side view of the chassis shows the oscillator coil, converter (V1) &
demodulator valve (V3), with the IF amplifier (V2) hidden.

siliconchip.com.au Australia's electronics magazine OCTOBER 2018 101

 light application of a heat gun. Don't
be tempted to use the sprays meant for
loosening bolts and screws. My expe-
rience with the Emerson 747 shows
that these lubricants can freeze the
adjustment slugs.
Luckily, in my case I was able to
remedy the problem by applying heat
but it's best to avoid the problem alto-
gether by not attempting to lubricate
coil slugs.
Also, the 747 service guide ad-
vises that you do not measure valve
filaments with an ohmmeter. Analog
meters can put out around 100mA on
low range and this advice also applies
to the 838.
Comparisons & performance
The most direct comparison I can
make is with Emerson's own 747, a
four-valve set similar in design to the
classic four-valve B7G portables of the
'50s and '60s.
Trimmer alignment is done with the chassis and batteries in place. C2 and C5
can then be adjusted by removing a small plate on the side of the case as shown. There's also the Hoffman "Nugget"
and the ingenious Crosley book radios,
earphone socket on this set. The lack of audible vibrator buzz was where the radio chassis nestles inside
8pF capacitor C19 bypasses the 45V most welcome. a "book-alike" case. Then there's the
HT supply from the B battery while As far as we can determine, no other contemporary all-transistor Regency
the LT supply is bypassed by 50pF hybrid radios were produced by Aus- TR1, as mentioned in the intro.
capacitor C21. tralian manufacturers although there The TR-1 used a hearing aid bat-
were a number of hybrid TV sets and tery that lasted only about 20 hours,
Editor's note here the situation was reversed: silicon compared to the Emerson 838 which I
At the end of the valve era, hybrid transistors did all the work in the small would estimate would last around 40
car radios were quite common as lo- signal stages, while valves were used hours, despite having a more power-
cal Australian manufacturers made the in the high voltage video and sweep ful output stage. So it compares quite
transition to transistors. stages (ie, yoke and EHT circuitry). favourably.
As with the American Emerson set Overall, I would have to say that the
described here, Australian manufac- Cleanup and adjustment 838 is a great performer for its size. Its
tured car radios used battery valves The example shown here was in audio output is adequate, and sensitiv-
for the RF sections and germanium good physical and electrical condition, ity is good — it's superior to many tran-
transistors in the audio stages, mostly needing only a polish to smarten it up. sistor sets of the day. The maximum
using a single germanium power tran- It worked right away and didn't need audio output is around 50mW. I did
sistor in Class-A mode. any adjustment. But if you do need to all my testing at 5mW as this seemed
The heavier current drain of the adjust an 838 (or its predecessor, the like a typical use case.
Class-A output stage was generally not 747), I have some helpful hints. I measured the sensitivity at 600kHz
a problem in these cases since the sets The chassis sits behind the front at around 30001/m, rising to 600pV/m
ran from the car's battery. cover. To gain access, remove the tun- at 1400kHz. In both cases, the signal-
These hybrid car radios were a sig- ing knob and gently prise the latch to-noise ratio was over 20dB. That
nificant advance on the earlier sets beside the tuning gang to begin releas- equates to around 900-180001/m at
with their vibrator power supplies. ing the front cover catches. Replace it a 50mW output, compared to adjusted
by first seating the catches at the op- figures for the TR-1 of 2000-2800pV/m
The Emerson posite end to the gang and then work for the same theoretical output level.
logo features
a take on towards it. Selectivity at —3dB measured
a G-clef The chassis needs to be removed for ±1.9kHz, at —60dB it was ±30kHz. The
followed by IF and oscillator core alignment. Trim- AGC allowed a 6dB increase in output
the phrase mer alignment must be done with the volume for a 60dB increase in input
"Emerson chassis in place in the cabinet, so an signal level. It was hard to overload,
Television access plate is provided for trimmer needing some 750mV/m before pro-
and Radio". capacitors C2 and C5 (see above). ducing noticeable distortion.
Be careful when adjusting the coil At 50mW audio output, Total Har-
slugs. Many sets of this era used a wax monic Distortion (THD) is around
seal and this is best eased off with the leo, with 6(3/0 THD at 40mW and only
102 SILICON CHIP Australia's electronics magazine siliconchip.com.au
 3% at 10mW. Audio response from
A brief history of Emerson volume control to the speaker (-6dB)
Vlc or Emersonincorporated a phonographcompany in 1915.Releasing America's firs is 300Hz to 6.5kHz, from antenna to
radio-phonograph combination in the 1930s, Emerson emerged from obscurity offering speaker is 280Hz to 1.7kHz.
the wildly successful "peewee" set in 1932. Small sets are notorious for having
With the peewee selling as many as 60% of all radios in the first half of 1933, Emer- a short battery life but this one draws
son's 50% share of this bonanza saw them become a major player. The 1947 release of a a modest 2.6mA from the B battery
10-inch television marked Emerson as an innovator, continuing to release the first clock (HT), falling to around 1.4mA on
radio, and solar-powered transistor pocket radio. strong stations.
The Emerson hybrid model 838 radio described in this article was released in 1955. This implies a life of more than 70
hours from the National Electronics
Distributors' Association (NEDA) Type
The miniaturisation of valves 213 battery, which had a typical capac-
The t • d ted around 1907 and the tetrode in 1919. By 1939, multi-functi ity of 140mAh.
valves (eg, diode-triodes) were common. That was also the year that the B7G series of I wasn't able to find data for the 4V
battery valves was released, which abandoned the historic pinch construction, connecting NEDA 1300 A battery. The set only
the internal assembly directly to a set of base pins embedded in the bottom sealing disc. draws about 50mA so I'm guessing
These valves were electrically similar to their older, octal predecessors but the B7G an original "A" battery would have a
series occupied some 25% of even the most compact octal valves' volumes. While the life of 40+ hours, as mentioned above,
B7G design allowed such advances as the revolutionary BC-611 "Handy-Talky", the pres- given that mercury batteries had ca-
sure for even greater miniaturisation remained. pacities roughly double that of alka-
Abandoning base pins entirely and bringing connecting wires through the envelope's line types.
base allowed further compaction. Three strategies emerged:
M. The E8 format has a cylindrical T3 (3/8-inch) envelope, retained a miniaturised Replacement batteries
version of the B7's base disc, but with eight connecting wire leads rather than pins. The The Eveready 415 45V battery (or
compaction was remarkable. The subminiature 1E8 valve has only 6.25% of the original its equivalents) can be bought online
6SA 7GT's volume. E8 types could be soldered directly in place or, with clipped leads, but at some $25+ it's an expensive way
plugged into sockets. to power these sets. I have previously
The E8 base also allowed the encapsulation, for example, of independent dual triodes, bundled up four 12V batteries (as used
a construction that had been impossible in B7G construction. Directly and indirectly- in remote controls) using everybody's
heated E8 valves were built, from VHF transmitting triodes to audio output pentodes, at favourite wrapping — duct tape. Like-
least one pentagrid, one triode-heptode and even a subminiature version of the iconic wise, I taped up three AA cells in se-
"Video Pentode", 6AC7. ries for the A battery.
H. A second approach reverted to pinch construction, with all leads (between three Bruce Wilkie (Radio Waves, January
and seven) in the one plane exiting through the flattened "press" at the base of the en- 2016) has a more elegant solution. His
velope. These types generally used a flattened envelope such as the T2X3 (2/8-inch x Crosley JM-8 "book radio" now uses a
3/8-inch). Some came with long "flying" leads and could be soldered in or (again with plastic AA holder for four 12V batter-
clipped leads) plugged into a socket. ies and a 3-cell holder from a cheap
W. A third class used a cylindrical envelope and base but presented the leads in a LED torch for three 1.5V cells. Bruce's
row, similar to the T2X3 and could also be soldered or plugged into sockets. A few EHT radio is very similar to my Emerson
rectifier diodes (designed for solid-state television sets) with two leads in the base and set and it's worth reading his article
one at the top (for the anode) used this construction to compare the two sets.
Generally, a reduction in filament/cathode heating power leads to a reduction in mutual I'd prefer to use three NiCd/NiMH
conductance and (at least for pentodes) in gain. The designers of the 1V6-1AH4-1AJ5- rechargeable cells (for about 3.8V total)
1AG4 series (replacing the 1R5-1T4-1S5-1S4), as used in the Emerson 838, economised to give closer to the original Mercury
a little by cutting filament currents from 50mA to 40mA. battery's 4V.
Although manufacturers managed to retain good performance in amplifying stages,
Emerson's designers still had to work hard when designing the 838 to ensure it was a Further reading
credible performer. • Emerson 838: siliconchip.com.
Miniaturisation and the cachet of "mili- au/link/aal7
tary-type" subminiature valves had appeal • Series-filament AGC systems, in
but the practically-minded would also be Radiotron Designer's Handbook (com-
wary of running costs, so battery life was plete, searchable PDF, around 90 MB
important too. in size): siliconchip.com.au/linkJaal8
The 20% reduction in filament current • Complete Centralab catalogue:
helped the 838 achieve a good battery life. www.audiophool.com/Techno.html
But the most significant improvement was (search for Centralab; it's a Deja View
from eliminating the typical Class-A valve [.DJVU] file, so you'll need the viewer
output stage and its poor efficiency, as plugin).
described in the main text. • Bruce Wilkie, The Crosley JM-8
Hybrid Book Novelty Radio, pp10-14,
Pentagrid converters from left to Radio Waves, Jan. 2016, Historical Ra-
right: 2A7, 6SA7, 1R5, 1E8 dio Society of Australia (HRSA). SC
L _J
siliconchip.com.au Australia's electronics magazine OCTOBER 2018 103