[John Byrns]

In article ,
(Don Pearce) wrote:

On Thu, 08 Sep 2011 09:31:41 -0500, John Byrns
wrote:

In article ,
(Don Pearce) wrote:

On Tue, 6 Sep 2011 09:03:54 -0700 (PDT), Engineer
wrote:

On Sep 5, 1:46*am, (Don Pearce) wrote:
On Sun, 4 Sep 2011 22:05:12 +0000, John L Stewart

wrote:

The drive requirements of the OP stage are not as critical as
mentioned
in earlier posts. The KT88, EL34 & 6550 all have G about double that
found in the original Mullard circuits where we find 6L6, 5881, Etc.

A single dominant pole is a safe way to go in this cct. There are a
couple of worked samples in RDH4 worth looking at.

Also strongly recommended by none other then Norman Crowhurst. You can
read about how to implement all in a couple of his articles.

I like the cct proposed by Alex Pogossov recently. The attachment
shows
how it is connected in an SE Amp. The frequency selective network is
in
the cathode of the first stage rather than between stages.

My opinion, anyway, having done it.

Cheers, John Stewart

PS- The bicycle went more than 1000 KM in August (622 miles).
Pedalling
again today. Hey Patrick, hows it going in OZ? * * * * * * * *J

Why would you want a circuit that drops the gain by about 16dB at
100Hz? What is its actual purpose?

Ans: To get the loop gain (forward gain times NFB gain) well below
unity at the low frequency where the phase lead becomes 180 degrees
(due to two RC couplings and one OPT.) Very low frequency oscillation
in the range 0.5 to 2 Hz can result if there in no low-frequency shelf
inside the loop.
Cheers,
Roger

So it is a horribly bodged solution to a problem that has no business
being there in the first place? The idea of implementing negative
feedback and then defeating it by killing the open loop gain it
mediates is beyond ludicrous.

On Sunday you suggested "a single pole that doesn't let up until beyond the
unity gain frequency", isn't that equally ludicrous as it also "kills the
open
loop gain"?

Not at all. It operates as a local feedback loop within the overall
loop. At lower frequencies, where it really matters to have
particularly the output stage included, the loop stretches from
beginning to end. As the frequency rises the loop transfers steadily
to just the voltage amplifier. This does result in a slight rise in
harmonic distortion above about 10kHz, but frankly by then who cares?

And this pole is going to have to be present in some or other form
anyway because high frequency instability needs to be tamed - so why
not make it simple, predictable and controllable?

So if I follow you, you are not talking about a single overall negative feedback
loop containing a dominant pole within, but instead you are talking about nested
feedback loops where the dominant pole within the overall feedback loop is
created by a second shorter internal feedback loop that rolls off the high
frequencies?

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/