On May 31, 1:09*pm, John Byrns wrote:
In article ,
*Patrick Turner wrote:
On May 29, 11:23*am, John Byrns wrote:
In article ,
*John L Stewart wrote:
Before this thread went off the rails did anyone mention that McIntosh
uses a seperate winding on the OPT for overall NFB?
A separate winding, a tertiary winding, provides better high frequency
stability
than taking the NFB from the secondary winding. *The down side of using a
tertiary winding for NFB is that the NFB doesn't include the secondary so
that
the damping factor suffers, this probably has no practical consequence, but
doesn't make for good ad copy.
Yes but the NFB winding position relative to anode turns or secondary
turns will have an effect on what is fed back.
Yes, the design/positioning of the tertiary winding affects the leakage
inductances between the various windings
I'd say that if a fine wire NFB winding is wound right over the top of
a thick wire speaker secondary then its signal is near equal to the
speaker's signal and the "error" made by the secondary Rw and LL and
phase shift et all is fed back for correction.
No tertiary can correct for the "Rw" of the secondary!
I'm not so sure. if the tertiary is wound close to anode coils, sure,
Sec Rw is not reduced by NFB. But if tertiary is wound close to Sec, a
drop in sec voltage gives a drop in tertiary voltage, and a FB signal
corrects the drop, no?
,
But if the FB tertiary is next to the anode primary then you have a
good hum free FB source with anode signal without LL and Sec Rw
effects and the amp is MUCH easier to stabilise unconditionally. Such
a low voltage and grounded winding gives a conveniently small FB
signal with low source resistance which can be applied to an earlier
stage very effectively. Such a signal picks up a sample of the whole
Va-a, and is a better NFB than using a cap plus R divider from one of
the PP anodes back to an input tube cathode. But bean counters mostly
hated paying for a dedicated NFB tertiary. Why have one when the
speaker secondary exists? Indeed why have one? if you make the OPT
with sufficient interleaving then the Williamson way of NFB is fine,
but bean counters hated Williamson too, and laughed at what he said
makers should do, so we all mainly ended up with OPT with very high LL
and appalling stability problems.
Marantz used an interesting NFB scheme that took the NFB from the secondary
at
low frequencies to provide a good damping factor and then used a crossover
filter to take the high frequency NFB from tertiary windings.
I'm not sure about that one. Some makers just used the speaker sec for
normal global NFB with R divider to feed V1 cathode but then the cap
that is normally strapped across the FB R to advance the phase of the
fed back signal at HF is not used, but a cap from V1 cathode is taken
to one of the OP tube anodes. I've always found no real joy in such
measures and I prefer the normal conventional Williamson idea where
one makes the OPT have wide BW which then pushes the regions where
instabilty occurs further below and above the AF band and thus to
where its easy to instal gain reducing and phase tailoring networks
while maintaining 10Hz to 65kHz full power bandwidth with complete
stability and 20dB global NFB is desired. I'll follow the Williamson
principles before I'll follow any others. I happen to like using
tertiary windings for NFB myself, but in the form of CFB windings
which are a part of the OPT primary and which carry Ia. Its the
Acoustical. Thus the OP stage is substantially linearised and given
the character of triode Ra and low THD while retaining the tertrode
AB1 or A1 power. Then only 10dB of global NFB is needed, if any at
all.
CFB is nice, but it does increase the demands on the driver stage substantially.
Depends how much CFB. Mcintosh amps need a max 150Vrms to 6550 output
tube grids for clipping. So the THD of the driver stage tugs at the
gains of so much CFB ( 50% ) in the OP stage.
Quad-II only have 10% CFB, and the KT66 are about as sensitive as a
50% UL stage. But the 10% CFB is more effective in getting Ra-a to the
same level as triodes. A 50% UL stage gives Ra-a about = to RLa-a,
several times the Ra-a of triode, broadly speaking.
I like 20% CFB, drive voltage goes to about 75Vrms to each 6550 grid
and I then build the drive amp to make far less THD than anything by
McIntosh or Quad, and voila, very low THD with very little global
NFB.
Patrick Turner.
d a
--
Regards,
John Byrns
Surf my web pages at, *http://fmamradios.com/- Hide quoted text -
- Show quoted text -- Hide quoted text -
- Show quoted text -