"Patrick Turner" wrote
...see below
Absolutely. Autotransformers are good for low ratios where isolation
is not important.
The cathode feedback, or more properly named distributed load in the
example you give, is another kettle of fish, of course. Dunno what
it's got to do with autotransformers though, except that you say
speaker windings in series...that then makes an autotransformer
between cathodes and speaker.
So then the speaker winding is *part* of the cathode winding..."in
series" doesn't quite capture the meaning IMO.
Interesting all the same.
So, with reference to my original point about wire thickness, how
would you deal with this in practise? Let's take an example. I use a
cathode windings of 10%, Zaa 5k and speaker 6ohms. Roughly
speaking, that gives me 3% speaker winding, a ratio of 3 to 1
compared to the cathode. So I could tap about a third of the
cathode windings to drive the speaker.
However, I require the speaker winding to carry several amps,
whereas the cathodes only require around 100mA. If I accomodate both
with the same wire necessary for the speaker, then there is more
leakage in the cathode circuit than would otherwise need to be the
case. If I use two windings of different wire gauge in series, part
of the cathode winding is thicker than need be, giving some penalty
in terms of leakage in the anode/cathode circuit.
My tendency has been in the opposite direction: to minimise a-to-k
leakage, at the expense of added capacitance, which I want anyway.
So my cathode windings are bifilar with the anodes'. I don't have a
bandwidth problem (about 16 to 45k without overall nfb), so the
autotransformer would not be worthwhile in my case, and defeat the
objective of the bifilar winding.
There is another issue he precisely where is best for the nfb to
come from. There are arguments for and against using the speaker
winding in the feedback loop. I prefer the simplicity of not doing
so, and suffer the slightly higher (but I believe less malevolent)
distortion as a result.
cheers, Ian
An auto transformer is useful to match awkward speaker
impedances to impedance sensitive amps.
To get a good match of a 2 ohm speaker, to say an amp which runs
best
with 8 ohms, a 2:1 step down ratio is required.
The impedance ratio is 4:1, so you get 8 ohms to 2 ohms.
To achive this transformation, with a typical low power tube amp,
which puts out only say 20 watts into 8 ohms, one could use a
toroidal power tranny, of around 300VA, with 2 x 40 volt windings
in series, and leave the mains primary unterminated.
Such a tranny will have a B of perhaps 1.2 Tesla at 40v at 50 Hz,
for
one 40v winding,
but at 20 watts into 8 ohms the voltage is only 12.6 vrms, so
with this voltage applied to two "40" volt windings,
the B at 50 Hz is only 0.189 Tesla, and at 12.5 Hz its still
only 0.756 Hz, and so still well away from saturation.
The BW is usually 100 kHz so the toroidal isn't a bad device to
couple speakers with, to get an impedance match at low losses,
and wide BW.
Various taps down the one winding can be placed to get
a wide range of impedance matches, so driving a woofer, midrange,
or a
tweeter
at any selected voltage level can be done can be done without
using
series
resistances, to achieve well equalised levels,
and the output impedance at the taps is always lower than the
amplifier
impedance,
unless one tries to step up the output voltage, which imho, is
pointless.
SO, those who do have 4 or lower Z speakers could easily
make a match to tube amps best suited to 16 ohms, such as Quad II,
even when set on the amp for 8 ohms.
There are plenty of tube amps which have the poor method
of impedance matches done by simply having a single secondary OPT
winding
for 16 ohms, with tap at 0.7 times the total turns for 8 ohms, and
a CT
for 4 ohms, which means that half the sec is not carrying speaker
current
when 4 ohms is used, and HF response is poorer.
The auto tranny at the speaker allows the amp to enjoy the 16 ohms
it
prefers,
and less OPT winding losses. The winding losses in the toroidal
auto tranny are minor.
The turns per volt required for a mains 300 VA tranny could be
about
2 turns per volt, so 80 turns are required fore 40v at mains F.
160 total winding turns would be fine for a 20 watt amp.
Its easy to place 160 turns of wire by hand around a toroidal
core,
using a shuttle made from a piece of old plastic tube, or broom
handle
with notches.
Use about 1.3 mm dia wire, and wind the right length for say 1/2
the
winding on the shuttle, then wind it off the shuttle onto the
toroid
and set the wires straight, tightly, and true, and neatly.
Tape insulation is easy to apply in lengths.
I have re-wound complete toroidal power
transformers like this, and of course it took ages to do the
thin wire high voltage turns, since there are so many turns.
The low voltage heater windings were dead easy to do, once I
made a sort of clamp to hold the partially wound tranny as I
placed on
the turns.
Using auto trannies to have the primary in series with a speaker
secondary
doesn't convey a huge advantage.
But in a PP amp, one could deliberately
split the primary into cathode and plate windings, a'la Quad
acoustical,
and have each half cathode winding in series with a grounded and
centre
tapped
speaker secondary, perhaps with other symetically placed taps to
allow
impedance matchings. The feedback effect is enhance this way.
The Quad OPTs don't allow for such elegant use of the speaker sec,
since it isn't a centre tapped winding.
The text books, including RDH4, will explain
the winding losses with two winding isolation trannies,
and with auto trannies, which are more efficient,
especially with a 2:1 step down transformer.
Where there is a will, there is a way.
Patrick Turner.
|