[HBR] Re: GB> RSGB One-Tube Keyer: UPDATE

[Walt Hutchens]

http://www.flickr.com/photos/15888272@N05/2415370027/

I had never seen this circuit and found it a bit confusing, but it's
actually not complicated.

V1a is a blocking oscillator. Initially there's no current flowing in
the tube and the cathode and grid are at the same voltage. When the
paddle switch is closed, the tube conducts. The cathode and grid
initially remain at the same voltage due to the 0.22 mfd cap across
the cathode resistor so the plate current is the zero bias current for
the plate resistor (set by the RATIO control) and plate voltage --
something around 150 volts.

The plate current begins to charge the 0.22 mfd cap and also flows in
T1, applying a positive voltage to the grid, relative to the cathode.
Brad's comment that the transformer wiring should checked is the first
thing that came to my mind as well.   I'd also check the resistance of
the various windings.

In theory, the positive voltage applied to the grid holds the tube in
saturation until a substantial voltage has built up across the cathode
resistor.  At that point, the current flowing in the tube (and hence
the transformer) decreases and that causes the transformer to push the
grid negative.  That causes even less current to flow, and the tube
shuts off regeneratively.

The voltage across the 0.22 mfd cap then decays through the SPEED
control and series resistor until the tube conducts again and the
cycle repeats.

I can see one possible fly in this ointment and it is the reason for
the "In theory ...," above. The time constant of the combination of
the weight control in the plate circuit and the 0.22 mfd cap would be
about that of (say) 100k and 0.22 mfd -- that is, about 1/50th of a
second and that's ignoring the plate resistance of the tube. The
response of the usual small output transformer at 50 CPS or less is
very low. Thus, the transformer will not deliver much voltage to the
grid. If there's not enough grid voltage the 0.22 mfd cap will simply
charge up until the tube current equals that flowing through the speed
control and series resistor. Then everything will stop until the
paddle is released.

You could try turning the RATIO control to near one end and pushing
the paddle that way. If it oscillates then, the transformer low
frequency response an issue.  But the 12AT7 plate resistance is most
of 100k, so a negative result doesn't eliminate this possibility.

One other thing I'd try is sticking a 470k resistor or so in series
with the connection from the transformer to the tube grid. That
eliminates the grid-cathode 'diode' as a load on the transformer,
meaning more energy stored (in the tx) and a better negative pulse
when the tube current starts to fall.

I would check the various resistances in the circuit and substitute a
couple of other 12AT7's -- even a slight heater-cathode short or bit
of gas in the tube will cause a problem, and those tubes often have
'em. 12AT7A's are most likely designed for series string operation
(most 'A' versions are) so they would be a better bet, but by no means
guaranteed.

Even the most minor transformer short would be fatal. Another useful
check would be hooking in another one with clip leads. Trying a better
quality transformer (better low frequency response) would help
eliminate that issue.

Heater-cathode issues are probably the reason for the use of a
negative supply rather than a positive one -- if the cathode was
positive relative to a grounded heater, you'd have a forward biased
diode across the cathode circuit, as well.

When transformer, tube, wiring, and parts value issues are eliminated,
I'd do voltage checks to get a better idea what's happening. If the
cathode does come up to some substantial voltage (like 50 volts or so) 
when the key is closed and then decays when it is released, then the 
problem is not enough feedback and the only question is 'why?'

I wonder if there could be an error in the diagram. But other than not
having a resistor in the blocking oscillator grid, I can't see
anything obvious.

Walt
KJ4KV