[Milsurplus] TCS Transmitter: OK Smart People; Why?

[Kenneth G. Gordon]

> > Those familiar with crystal-controlled transmitters of the 40s 
> > and 50s will recognized this pentode oscillator circuit:
> >
> > http://home.netcom.com/~arc5/TCS/TCSXtal.jpg

That is the standard "grid-plate" crystal oscillator circuit, and I think I can tell 
you why it isn't a good crystal oscillator: see below.

> > ...The TCS values bias harder and reduce the screen voltage considerably. As a
> > result, crystals like FT-243, CR-1, HC-6 and even large-blank FT-241 will not
> > reliably function or fail to oscillate altogether.  However- I have a couple
> > of pre-war, Collins-made large-blank transmit crystals:
>
> That seems improbable to me,

Me too.

> because Collins crystals apparently often did NOT
> work.  Collins and several contractors made the nt-40068 crystals.  Hamilton
> manuals also refers to nt-40130.
> 
> There is an inherent design problem that can occur even when using the proper
> crystal unit.  Section 4.6.1.1 in the manual for Collins-made sets discusses CW
> "keying problems" that can result when USING CRYSTAL CONTROL.  Keying will
> likely be  unsatisfactory above 20 wpm, and some crystals will cause
> unsatisfactory keying at any speed.  The manual states that this is inherent to
> the crystal and is not the fault of the transmitter.
> 
> That's a shabby way to address a real problem!

Absolutely agree.

As I see it, there are at least two problems inherent in the above schematic: 
1) the 12A6 is a beam tube and 2) feed-back in such circuits is controlled 
more by the capacitance between the SCREEN and ground than by almost 
any other path in the circuit.

>From the RSGB handbook of some time ago, beam-tubes do NOT make 
good oscillators of any kind, especially ECOs, due to the fact that the 
beam-forming plates (which are NOT a true screen grid, BTW) couples the 
RF energy from the grid-cathode-screen oscillator directly to the plate, 
bypassing any ECO shielding effect.

Despite this, many circuits of the WWII era used beam tubes of one type or 
another as oscillators. The TCS with its 12A6, the BC-610 with its 6V6 or 
6L6, the Paraset with its 6V6, and others which used tubes up to and 
including the 807.

The U.S. Navy, for some reason, was far more careful about this issue, using 
tubes like the 837, and other such "real" tetrodes or pentodes but 
NON-beam tubes as oscillators.

If you cannot find another tube with a 12.6 V filament at 150 mA with the 
same base diagram as the 12A6 to plug in, I would recommend a 6AG7 with 
a 10 ohm 5 watt resistor in series with its filament. 

The trouble with this is that since the 6AG7 requires 650 mA filament current, 
and if the TCS uses the 12A6's filament in series with some other tube in 
order to get to 25.2 V, then that would unbalance your filament string.

You would then have to compensate for this by paralleling the other tube 
with an appropriate resistor.

Another possibility is to build a socket adapter and plug a 
12BY7/12DQ7/12GN7 into it, although that tube requires 300 mA filament 
current, so you would have to parallel the "other" tube's filament with a 
resistor to balance the filament currents.

Another excellent tube for this purpose would be the 6CL6, which is, 
essentially, a 9-pin 6AG7.

Lastly, after you have changed the tube to something far better, add a small 
mica cap between the screen connection and ground. Start with something 
like 25 pfd, and adjust to get best keying.

Another very interesting fact I gleaned from my old RSGB handbook is that 
any pentode, when used as an ECO, exhibits a very pronounced 
compensation for drift if the screen and plate voltages are set to the proper 
ratio. 

This is because changes in screen voltage cause an oscillator to drift in one 
direction while changes in plate voltage cause an oscillator to drift in the 
opposite direction (frequency-wise). Therefore there is generally a certain 
ratio of screen to plate voltages which automatically compensates for drift 
caused by voltage changes.

Ken W7EKB