[Collins] More re: 6146/A vs 6146B/W

[Dr. Gerald N. Johnson, electrical engineer]

I've done a little digging. I downloaded the latest manual for the KWM-2
from the CCA page (nothing there for 32S3) and the parts list. Section 7
has an extensive section on updates and variations over the years. It
mentions changes to capacitor values in the RF feedback divider, but not
the changes in the PA neutralization circuit. Its worth reading though a
12 MB download on a 28.8 dial up takes lots of patience. Be sure to save
it first.

The 6146B, having greater power dissipation, has different
interelectrode capacitances just because there's more and larger
hardware inside. Better shielding could decrease feedback capacitance.

There has been a change in the PA neutralization circuit from the
earliest (as shown in the yellow book on SSB) to the latest. The early
neutralization circuit used a 10 pf 5 KV Centralab 858 series capacitor
in series with an 8-50 PF ceramic variable capacitor. That gives an
effective range of 4.44 to 8.33 pf. There is a 100K resistor to ground
from the junction between the two parts. As a resistor it has no effect,
though it will add a bit of C to ground and that T network will have a
lower effective capacitance range. I measured a 100K 1/2 watt carbon
composition resistor as having 0.4 pf capacitance. Neglecting the
resistance, only using its capacitance I compute a capacitance range of
4.35 to 8.28 pf. Certainly the two ceramic capacitors will be somewhat
temperature sensitive, and the greater the setting the more the RF
current for heating.

The new circuit uses a 1000 pf 2 KV ceramic, a 470K 1 watt resistor to
ground (has full RF plate swing applied), and a 1.4 to 8.1 pf 1250 volt
variable capacitor. 1250 volt tiny variable capacitors are no more stock
these days than the 8-50 ceramic. The 0.4 pf of the resistor 470K has no
effect at all (just shunts the PA tuning) so the effective C range is
1.4 to 8.1 pf. Better on the low side than the old circuit, the same as
the old circuit on the high C side neglecting that shunt capacitance of
the resistor and its mounting.

It would be possible to pad either the 10 pf (say with 2 or 3 pf with at
least a 2K voltage rating) or the ceramic variable (with maybe 18 pf) to
increase the maximum C range. The RF current in the neutralization
circuit would increase leading to more drift of the capacitor not
padded.

In the later circuit it would be practical to pad the air trimmer with 2
or 3 pt to add that much to the maximum capacity range, though any
padding capacitor may add to the temperature drift.

It should be possible to start with a 25 or 30 pf 500 volt variable
capacitor and pull plates to make an 8 pf 1250 volt. You have to
essentially double all the air spaces (plus the thickness of the plates
removed). Starting with a little larger capacitor should allow a little
larger neutralization capacitance range.

The effective range of the air space neutralization capacitor will also
depend on precisely where its mounted whether the shaft or the plates
are in the plate compartment or in the grid compartment. And if the
shaft is in the plate compartment, which side of the circuit the shaft
is connected to, if it has a shaft.

It might ba as effective as major circuit modifications to connect a
piece of #16 buss wire to the junction of the two capacitors in the old
circuit or the junction of the air capacitor and the grid circuit bypass
and run it through a convenient hole into the plate compartment and
adjust how long it is and how close to a PA tube to add a few PF of low
loss capacitance.

If I needed a little more neutralization C in either circuit, I'd be
inclined to use the buss wire (insulated at critical places with Teflon
sleeving). Many a tetrode and pentode PA has been neutralized that way
at frequencies from HF through UHF.

73, Jerry, K0CQ, Technical Advisor to the CRA