WSQT wrote:

You will probably be best off winding your own. My advice is this:
find the core size required to handle an amount of power at 60HZ equal
to the audio power needed, find an old power transformer of that core
size, and unwind it. Count the turns on the 117VAC winding to get turns
per volt, and figure the wire size from the DC input to the modulator
for the primary windings, and from the DC input to the PA(for each
side), and turns from the turns per volt for power. Do not push wire
size the the limit(small) becuase the windings also must handle the
peak currents produced by the audio power, which are higher than the DC
average.

This transformer will have more than enough iron in the core to handle
the AC power, as audio is at a higher frequency and the inductive
reactance higher. DC in the secondary will lower the inductance, but
you should still have enough. If you were trying to handle deep bass
frequencies(not needed nor wanted in a "communications" app as they eat
percentage of modulation), you would use a still larger core, figuring
audio power PLUS the power represented by a 60HZ AC current whose RMS
value is equal to the current drawn by the PA. This "overkill"
transformer would give less distortion on deep bass.

For a ham or any other communications application, frequencies below
300 HZ or so are far less important than they are to a broadcaster, but
be SURE not to allow frequencies to pass the modulator's driver that
the output transformer is not efficient at, or heavy currents will be
drawn and efficiency will suffer badly. If you design a modulation
transformer only to handle a narrowband voice range, the modulator
input transformer should also reject any frequencies lower than the
output transformer can handle, or these frequencies can be filtered out
elsewhere.

Otherwise, just speaking too close to the mike(which produces a deep
bass response) will heat up your modulator.


A hint on rewinding transformers. When you disassemble a power transformer
you will note that the laminations alternate (IE: open end of even numbered "E"s
point left, odd numbered ones right). If you took apart a filter choke you
would find that ALL the laminations were oriented the same way. For an
audio transformer, arrange the laminations in groups, and alternate the
groups. Divide the total laminations into 3-10 groups.

Why? In the case of the choke, which must carry DC, we don't want the magnetic
flux to saturate or the inductance will start to drop at some current.
Having all the laminations go in one way leaves a nice air gap in the core
which prevents core saturation. In the case of a power transformer which
carries only AC we want the magnetic flux to be tight to keep efficiency up.
For an audio transformer which must carry some dc (tube plate current),
and ac (audio), a mix of these works out best. Note that in the case of
a push pull primary (modulation or output transformer) the balanced dc
current cancels out and we don't get much saturation of the core. A single
ended output transformer WILL have this problem, as will the secondary
winding of the modulation transformer.

A swinging filter choke should have some core saturation, probably it uses
two or three groups of laminations.