RCA: Hollywood Star, Model 32263

Apparently I can't help myself, I picked up this highly impractical but cool-looking projector with a matching speaker.  It's an RCA that was made under licence in England.  

Being an English amp, it's running EL84s in the power stage, and quite fortunately, it was full of original Mullard valves - all in great condition, with lots of life left.  

RCA Hollywood Star Projector Amplifier

The signal path is:

EF86 pentode → volume → triode-connected EF86 → James tone controls → ECC83 driver and phase inverter → cathode biased EL84s

The schematic above has been simplified by removing all projector-related components, along with the projector’s mains wiring and switching.  The schematic is also missing a couple of components that appear to have been factory-fitted.  These included a 330pF capacitor from the phase-inverter input grid to ground (pin 7, the boost stage) and a 33kΩ grid-stopper. 

EF86 preamplifier

The first EF86 operates as a high-gain pentode. It has a 100 kΩ anode resistor and a fully bypassed 1 kΩ cathode resistor. Its screen is supplied through 390 kΩ and bypassed to the cathode.

The input impedance is approximately 680 kΩ, set by the grid-leak resistor. A 5 nF coupling capacitor feeds the 1 MΩ volume control, trimming some bass before the following stages.  A 150 pF bright capacitor preserves treble at lower volume settings.

Triode-connected EF86

The second EF86 has its screen connected to its anode, making it operate as a triode.  This gives less gain than pentode operation, and a lower output impedance—roughly 15 kΩ—which makes it well suited to driving the passive tone controls.

It uses a 100 kΩ anode resistor and a bypassed 2.2 kΩ cathode resistor. A 10 nF capacitor couples it to the tone network.

James tone controls

The RCA uses passive James bass and treble controls

The main values are:

• 1 MΩ bass and treble controls

• 1 MΩ and 100 kΩ bass resistors

• 500 pF and 5 nF bass capacitors

• 150 pF and 1.5 nF treble capacitors

These values place the broad centre or pivot region at approximately 320–330 Hz. This is not a sharp cutoff: below this region the bass control becomes increasingly effective, while the treble control has progressively more influence above it.

12AX7 / ECC83 driver and phase inverter

The first half of the ECC83 restores signal lost in the passive tone network. It uses a 220 kΩ anode resistor and an unbypassed 3.3 kΩ cathode resistor. Global negative feedback from the output transformer also returns to this cathode.

The second half is a concertina phase inverter. Equal 100 kΩ anode and cathode loads produce two approximately equal signals of opposite phase for the EL84 output valves.

Each EL84 is driven through a 5 nF coupling capacitor, 10 kΩ grid stopper and 470 kΩ grid-leak resistor. The relatively small coupling capacitors also limit excessive bass entering the output stage.

EL84 output stage

The output stage uses two fixed-bias EL84s in push-pull - cathode biased and bypassed.  DC from the cathode is used to elevate the heaters.

Negative feedback

Feedback is taken from output-transformer secondary pin 4, with pin 3 grounded. It returns to the ECC83 driver cathode through a 100 kΩ resistor.

A 33 pF capacitor in parallel with this resistor mainly affects very high frequencies and was probably included to maintain feedback-loop stability.

Speaker cab

Single 10" alnico speaker, measures ~11DC resistance - everything points to it being a Goodmans.  The cab is cool, but a tad flimsy.   Sounds pretty nice, great mids and highs, which seems common with these.  I had to reglue some plywood that was separating.

Notes and modifications

• B+ is now approximately 320 V, reduced from around 335 V.

• R28, the shared EL84 cathode-bias resistor, was increased from 130 Ω to 180 Ω because the output valves were running well above a reasonable dissipation level.

• The C10A 150 pF bright capacitor across the volume control was removed.

• R9, the first EF86 cathode resistor, was increased from 1 kΩ to 2.7 kΩ. The original stage had excessive gain and was distorting prematurely; the higher resistance reduced the valve current and improved its bias and input headroom.

• R15 and R16 in the bass-control network were replaced. Both had drifted more than 50% above their specified values, shifting the tone-control response too low and substantially reducing the useful range of the bass control.

• R18, a 220 kΩ anode resistor, was replaced after it had drifted to approximately 275 kΩ.

• An additional power-supply filter stage was installed, consisting of a 150 Ω dropping resistor and 33u filter capacitor. This reduced the B+ voltage and provided additional smoothing.

• An undocumented 330pF capacitor from pin 7 of the 12AX7 phase inverter to ground was removed.  The 33 kΩ grid stopper was retained.

• The unused magnetic/optical sound selector switch was repurposed as a negative-feedback on/off switch.

RCA: BA-31C Preamp, Vero Layout (The Balmoral Broadcast)

The RCA BA-31C would have been found in broadcast consoles of the day (late sixties), as a microphone preamp module.  

I think it may have been the starting point for the Hudson Broadcast, as they look quite similar and I've heard some references made to RCA.  If it is what they used, I can see why they did - it's very tweakable, with some nice options for setting gain as well as high and low-end roll-off.   

The original runs at +/-30V.   I was aiming for a 9v single rail supply to keep it simple, and of course without expensive transformers.

RCA BA-31C MICROPHONE PREAMP - ORIGINAL SCHEMATIC

RCA BA-31C - MODDED FOR GUITAR

I tried to keep things simple, while still maintaining the structure of the circuit.  

Note:   This is still a work in progress, still some fine tuning and changes in progress.

GAIN

Like a Hudson Broadcast, it runs from clean boost to an overdriven fuzz.  It's not super fuzzy like a fuzz face, but you can hear it.

After experimenting on the breadboard, I decided to use an A1K pot for R6, and set R10 to 22k - 68k (anything a bit bigger than the original 6.8k resistor is fine, so long as it still has some feedback).  Decreasing R5 also provides a bit more dirt.  

FREQUENCY RESPONSE

With a few capacitor value changes the frequency response is easily adjusted.  

Increase C5 to roll-off high frequencies, although this is only needed at the highest gain settings, on lower gain settings, the tops are already reduced.

Decrease C3 to roll-off lows after Q1.  1u is pretty flat, can go down to relatively small values (22n as an example).  

MODDED RCA BA-31C ON THE BREADBOARD / SCOPE

So it works....  this is using a B5k pot for R6, and I think a 100n on the output as it was sitting right in front of me at the time.  Will change over to 1u.

PNP is a germanium 2SB457, NPN is a BC108.   It really isn't at all fussy about what germanium is used for Q2. 

 

caution: this starts at low volume and increases - watch your ears / speakers.

VERO LAYOUT - THE BALMORAL BROADCAST

Initially I thought I would end up with a few switches for bass and treble - but for now I decided to keep it simple.  Think I will most likely land on gain, volume, and a fat switch (which will change the coupling cap value between Q1 and 2).  For me it doesn't get bright enough to need control on the treble - if anything it's a bit low on treble on cleaner settings.

At low gain settings, it has a slightly rolled off high-end, with is nice and bluesy, maybe a bit of a jazz tone.  I did try it with a boost in front, which works pretty well, although it can use a bright cap across the boost level control.

This layout is a little different to the schematic above, as I decided to change the output cap.  I've called it the Balmoral Broadcast - due to some similarities to the Hudson, and I live in Balmoral.

 

Note:  not final versions