Thread: Feedback
View Single Post
  #14   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Feedback



Sander deWaal wrote:

John Byrns said:

A cathode resistor is local feedback, and some even argue that a
triode has inherent feedback.



Strictly speaking that's an *unbypassed* cathode resistor but I know
what you're driving at.


That's what I meant, sorry.
It's njust not a habit of mine to bypass cathodes ;-)


I know of no (commercial) transistor amp that can work without loop
feedback.



It could be done though. It's simply not conventional to do so.


Wasn't it done about a year ago right here in this group, although it
wasn't commercial which I assume is the reason for the disclaimer.


Yup. I've tried it with BJTs, and even my hybrid (MOSFET out) amps
don't use global feedback for AC (there is a DC servo loop, though).

But commercially, I have not seen it.
Densen claimed they did it, but it turned out there *was* global
feedback in there..........

But we're digressing into silicon again, gentlemen, some readers will
punish us for that ;-)


There are plenty of people who have used only emitter or source follower
connected output stages which is a localised application of perhaps
anything from 15 to 50 dB of series voltage NFB.

Some of these ppl then try to jump on the bandwagon to gain approval and
sales with ppl who hate global NFB.
By not including global NFB, they try to con us into thinking there isn't
any NFB, but
we know the truth here.

Now providing the output stages are in class A, then the local follower NFB
is enough to render
the silicon output stage to be as linear as any triode output stage of the
same power.
Typically, a single complememntary source follower couple of NPN and PNP
power mosfets
can give less than 1% THD into 16 ohms at 25 watts at clipping in pure class
A.
The THD is a combination of mainly 2H and 3H and fairly benign THD at low
levels where
such an amp would typically be used to make less than a watt of average
power,
and where the THD is typically less than 0.2%.
Using 6 mosfets instead of 2 gives 1/3 of the THD because the load each
mosfet sees in class A rises from
32 ohms to 96 ohms and the amount of follower FB is increased 3 times.
Bias can be reduced to keep the sixpack cool, and the AB transition is
tolerable.

But you have to provide a drive voltage slightly greater than the speaker
voltage.

So the driver amp using BJTs has to be linear, and a ziclai pair with its
own NFB
and a +70V supply is a nice candidate, or else a tubed driver stage
with a couple of triodes and FB loop.

The simplest driver would be say a pair of EL84 in triode in parallel
with choke loading and this should give 20Vrms at less than 0.5%, or a lot
less than the output
stage of mosfets, and the Rout of the triode drive stage will be only about
1k, have
gain at about 18, have no NFB and give sufficient bandwidth in the face of
the
input capacitance of the mosfets which is reduced from the Cg-s stated in
the data due to the
follower connection.

Better results are obtained when the mosfet amp uses an OPT and is driven
with a PP
LTP because the output device symmetry acts to keep the THD much lower than
N and P devices.
The OPT allows load matching.
See my page on class A mosfets with a combination of shunt NFB and global FB
at
http://www.turneraudio.com.au/solids...no-mosfet.html

My intended improvement to my ciruitry is to replace the bjts at the input
with
2SK369 high gm j-fets which then should drive the mosfet output stage
without the bjt N&D of the present set up.

In case you were not aware of what simplicity really was, try having a look
at
the ZEUS amplifier at Sue Parker's website at
http://www.susan-parker.co.uk/zeus.htm

The only problem for us tubies with Sue's amp is the low input resistance
because of the
step up input tranny.

Hence a PP triode LTP driving the mosfet gates should make a lot of sense to
those of us
who are not fond of input trannies, even when they are wide band toroidals.

I am not worried by people wanting yo lynch me on a stout oak tree branch
because
i dare to recommend silicon as a means to an end.

I like blondes, brunettes, AND redheads.

Patrick Turner.





--
"Due knot trussed yore spell chequer two fined awl miss steaks."