[Amps] Instability update, Apr 16

[N1BUG]

I will no doubt say at least one incorrect/stupid thing in my 
response to Bill. I trust and hope someone will point it out so I 
can learn from it!

On 04/16/2013 11:38 PM, Bill Turner wrote:
> Your tube may be oscillating as an old fashioned TPTG (Tuned Plate Tuned
> Grid) oscillator.

That is my present theory, based partly on the fact the frequency of 
oscillation varies smoothly over quite a range as the plate pi 
network loading capacitance is varied, and to a somewhat lesser 
degree when the plate tune capacitance is varied. I suspect somehow 
the plate circuit is resonating (albeit probably with some 
ridiculous Q) at LF. I would add the amplitude of the parasitic 
oscillation varies also, becoming greater as the capacitances are 
increased. This isn't an erratic or jumpy thing at all. You can see 
the frequency and amplitude of oscillation smoothly vary, and even 
gradually come to a stop as the plate circuit capacitances are 
reduced beyond some point. I have some experience with an 
intentional TPTG oscillator, and this behaves very much like it!

> Something in the plate circuit is present a large impedance at LF. Do you
> have a relatively large choke such as a 2.5 mHy or so in the plate lead as a
> filter? If so, try removing it.

I don't. The plate choke is a knock-off of the one that was (is?) in 
the ARRL Handbooks. It should be a few hundred uHy. There is a 4700 
pF bypass cap at the cold end of that choke, then a heavy conductor 
straight to the power supply (where there is a 50 ohm 50 watt glitch 
resistor in series).

The schematic I posted yesterday shows additional parts which are 
*not* present in this HF amp (C10, RFC3, C18)

http://www.n1bug.com/tech/4CX1500B-6m/6m4CX1500Bsch-hires.jpg

> If the impedance
> is low enough, the tube will not oscillate. The same statement holds true
> for the grid-cathode circuit.

The grid-cathode circuit worries me.

C2, the grid bypass capacitor, has a reactance of 60 ohms or more at 
the frequency of the LF parasitic. R2 is 150 ohms and C8 has a 
reactance of 350 ohms or more. I say "or more" because the frequency 
of the parasitic varies with plate pi network adjustment as noted above.

C6 and C7 have a reactance of about 350 ohms or more at the 
frequency of the LF parasitic.

I have no idea what the reactance of the filament choke, RFC1, is.

I wonder if some combination of RFC1 and the grid/cathode bypass 
caps is forming a resonant circuit at LF. Or perhaps the grid just 
isn't bypassed well enough at these low frequencies. My next step is 
to try increasing capacitance of C2, C6, C7, and C8 to see if that 
has any affect. I don't know what else to try at this point.

I recall many years ago I had a 144 MHz amp using a 4CX1000A with a 
very similar circuit configuration except that the input to the 
cathode was not loaded by a resistor and was tuned with an L network 
(actually a pi network when you include the tube input capacitance). 
It has been so long I have forgotten the symptoms, but I remember 
something odd was happening with that amp. I distinctly recall 
"curing" it by placing .1 and 1 uF (!?) caps in parallel with the 
.001 uF feethroughs which were the equivalent of C6, C7, C8 in this 
HF amp I'm fighting with now. I have no idea what led me to try that 
at the time. I wish I remembered more details about that problem.

73

-- 
Paul Kelley, N1BUG
RFI Committee chair,
Piscataquis Amateur Radio Club
http://www.k1pq.org