[Collins] 75S-3B audio output tube

[Dr. Gerald N. Johnson]

On Mon, 2008-12-15 at 18:43 -0600, Larry WA9VRH wrote:
> Hi Jerry,
> 
> Please let me know what you find on the 51S-1. I have one that I would like 
> to do this to.
> 
> Glad I could find it.
> 73 Larry WA9VRH
> 
Simply said, the K5UGM mod is to reduce the grid bias by putting a 330k
from grid to ground. Then add some cathode bias back by ungrounding the
cathode with a 47 ohm resistor bypassed with a 25 uf capacitor.

It will work. But all the receivers are not the same grid resistance,
cathode resistance, plate voltage or screen voltage.

I analyzed the operations based on the receiver manuals, including their
parts list, normal voltages chart, and schematic, plus tube data sheets
from General Electric and others.

Actually, simply plugging a 6AQ5 in the 6BF5 socket will work in the 75S
receivers so long as they are running on AC or external 6 volts. Where
the heaters are run in series parallel as in the 75S receivers and the
51S-1() the few tubes in parallel with the 6BF5 socket (now holding a
6AQ5) will be seeing a higher heater voltage that will shorten their
lives, probably seriously. And the rest of the receiver's tubes will be
running low voltage making them seem to work less well with time.

The simple change is not optimum. First off, the 6AQ5 doesn't need as
much bias voltage as the 6BF5 so it will be less linear showing
considerable second harmonic distortion. Secondly the gain will be lower
because the transconductance of the 6AQ5 is half that of the 6BF5. Third
the power output will be lower. How much the power output will be lower
depends on the model and MCN version of the receiver. My books show
sometimes different values on the schematic, the voltages chart, and the
parts list and for different editions of the books.

The 6AQ5 would prefer a higher plate voltage and in any case a 5K load
rather than the 2.5K load in the 75S receivers. The 6AQ5 is rated at a
significantly higher plate dissipation value, 12 watts vs 5.5 watts for
the 6BF5.

I compute from the tube curve and supply voltage that in the 'S-1
through S-3A, with 16 volts peak to peak audio on the grid, the 6BF5
produced 1.47 watts. The 6AQ5 produces 0.287 watt under the same
conditions with 11 volts of grid bias (circuit not changed). With the
6AQ5 bias reduced to 8 volts (as in the K5UGM circuit) the power out
with 16 volts peak to peak drive is 0.43 watt, down 5.3 dB from the
maximum power out of the 6BF5 and at 8 volts of bias the 6AQ5 distortion
is less.

Were I to use a 6AQ5 in the 'S-1 through S-3A, I would probably add a
shunt resistor grid to ground to achieve -8 volts on the grid, which
computes to be a shunt resistor of 620K. Though I don't object to the
cathode resistor, it makes changing tubes less critical, and if not
bypassed reduces the gain further while significantly improving the
linearity from the inverse feedback. Changing the total grid bias from
-11 volts to -8 also significantly improves the linearity of the 6AQ5 at
the cost of heat by raising the plate current to something around 32
milliamps.

In the 'S3B/C, there may be a 27 ohm unbypassed cathode resistor and the
grid resistor might be 220K or 330K. Plate and screen voltages are
higher. I compute that the 6BF5 will put out 1.54 watts of audio while
dissipating 4.47 watts in the plate. The 6AQ5 with no circuit mod will
put out 0.33 watt (both with 16 volts peak to peak grid drive), down 6.7
dB. With the 6AQ5 grid bias reduced to -8 volts (grid to cathode), the
plate current will be 35 milliamps and the plate dissipation 6.48 watts
and the power output will be 0.48 watt, down 5 dB.

If the 3B/C grid resistor is 220K, and the cathode resistor 27 ohms,
330K grid to ground is an appropriate and simple mod. The 6AQ5 is
dissipating 2 more watts in the plate than the 6BF5 but well within its
ratings while saving 4.7 watts of heater power but producing 1/3 the
audio power for the same drive.

The power output of the 6AQ5 can be helped by using an 8 ohm speaker
instead of a four ohm speaker at the cost of output transformer
bandwidth. The capacitor shunting the plate will have a lower frequency
roll off, possibly making the audio sound muffled. That capacitor has
been unhooked in my receiver for probably 40 years because it rolled off
2975 RTTY high tone too much. Cutting that capacitor in half would help
this situation. The transformer will also raise the LF roll off from
insufficient primary inductance with the 8 ohm load. 

In summary: The 6AQ5 mod is practical at a cost of at least 5 dB lower
gain and power output. Its best when the grid to cathode voltage is
adjusted to -8 volts (grid negative) with any combination of resistor to
ground at the grid and a 27 or 47 ohm cathode resistor. It shouldn't be
used when the heaters are in a series parallel string, whether 75S
receiver operating from external 12 or 24 volts or a 51S-1(). Despite
its heat, the original 6BF5 will produce more audio with less drive.

Its important to be sure the bias on the 6BF5 is still a total of -11
volts, and there is a small electrolytic in the bias divider that if it
shorts will reduce or eliminate that grid bias frying the 6BF5. That is
C29, positive side grounded.

-- 
73, Jerry, K0CQ, Technical Advisor to the CRA
All content copyright Dr. Gerald N. Johnson, electrical engineer