As per a prior thread about the Mullard 5-20, I was looking for a suitable
circuit to build and use a pair of Dynaco 6.6K PP OPTs I recently acquired.

I listed the weaknesses of the Mullard 5-20 amp that Morgan Jones cited in
his book, "Valve Amplifiers." However, I made an error when I posted a
Hafler modified version of the circuit; Jones was referring to the original
(unmodified) version.

Nonetheless, I just posted a .pdf schematic on
alt.binaries.schematics.electronic (entitled Dynaco A420 Project) that was
designed by Hafler & Keroes in 1951. It is a justifiably famous circuit,
for the use ultralinear and a nicely balanced topology.

Apparently, V1 is a differential phase splitter with gain? I don't
understand the operation of V1b . . . Why would the grid be connected to the
junction of R6 & R7? It looks to me like the output of either V1a or V1b
would appear at the grid of V1b, and why doesn't it oscillate?

I like the concept of the mu follower, and amps that I have built using it
sound better than ones that don't. I could see how one could be used with
V1a, but I'm totally clueless about V1b. If a mu follower could be
employed, might a 6SN7 be used for V1 instead of the higher mu 6SL7?

Maybe I should just leave well enough alone and implement the original
circuit . . . .

Sorry to return once again to the subject of tubes, but I think this circuit
is interesting and would love to know more about how it works.

Tia,

Jon

(a.k.a. "the dishonest garage trader, to which Andre shall add, the
"ignorant" dishonest garage trader . . . . ;-)

"Jon Yaeger" wrote in message
...
As per a prior thread about the Mullard 5-20, I was looking for a suitable
circuit to build and use a pair of Dynaco 6.6K PP OPTs I recently
acquired.

I listed the weaknesses of the Mullard 5-20 amp that Morgan Jones cited
in
his book, "Valve Amplifiers." However, I made an error when I posted a
Hafler modified version of the circuit; Jones was referring to the
original
(unmodified) version.

Nonetheless, I just posted a .pdf schematic on
alt.binaries.schematics.electronic (entitled Dynaco A420 Project) that was
designed by Hafler & Keroes in 1951. It is a justifiably famous circuit,
for the use ultralinear and a nicely balanced topology.

Apparently, V1 is a differential phase splitter with gain? I don't
understand the operation of V1b . . . Why would the grid be connected to
the
junction of R6 & R7? It looks to me like the output of either V1a or V1b
would appear at the grid of V1b, and why doesn't it oscillate?

I like the concept of the mu follower, and amps that I have built using it
sound better than ones that don't. I could see how one could be used with
V1a, but I'm totally clueless about V1b. If a mu follower could be
employed, might a 6SN7 be used for V1 instead of the higher mu 6SL7?

Maybe I should just leave well enough alone and implement the original
circuit . . . .

Sorry to return once again to the subject of tubes, but I think this
circuit
is interesting and would love to know more about how it works.

Tia,

Jon

(a.k.a. "the dishonest garage trader, to which Andre shall add, the
"ignorant" dishonest garage trader . . . . ;-)


Hi John -
Once again, it's just drawn weird. Forget about R2 - that's just there for
the global FB. Just connect V1A & V1B cats together. Now the grid of V1a
sees the input, and sees ground through a R1 grid leak. With no signal,
it's at ground. The grid of V1b sees the ground through R7 & R9, it's at
ground with no signal too - at least we've got the things drawing the same
current at 0 signal.
Now look at R6 & R7 - a voltage divider, feedin' the grid of V1b. It's
not seeing any B+, 'cos of the coupling caps C2& C3 - only the AC component
of the difference between the plates of V1A & V1b.
Let's drive the grid of V1a positive - 1a will "pull" it's plate down ,
driving the grid of 1b down, (negative), letting the plate of 1b "float up",
thus doing what a phase inverter's supposed to do. As the plate of 1B
floats up, itraises the voltge on it's grid, but, since R6 is smaller than
R7, the gridis less effected by the voltage on it's own plate than by the
voltage of 1a's plate - the grid is more affected by 1a's plate than by it's
own (which prevents the grid from following the plate of 1a & NFB &
stuff...) Lordy, could I be more unclear.
Hope this helps some - just try looking at the circuit in extreme states -
input all the way positive, input 0 input all the way negative, and forget
the R2 & everything past R8 & R9, to make things simpler. You'll see that
the values R6 & R7 don't have to be too precise to get this inverter to be
balanced...
-dim

Jon Yaeger wrote:

As per a prior thread about the Mullard 5-20, I was looking for a suitable
circuit to build and use a pair of Dynaco 6.6K PP OPTs I recently acquired.

I listed the weaknesses of the Mullard 5-20 amp that Morgan Jones cited in
his book, "Valve Amplifiers." However, I made an error when I posted a
Hafler modified version of the circuit; Jones was referring to the original
(unmodified) version.

Nonetheless, I just posted a .pdf schematic on
alt.binaries.schematics.electronic (entitled Dynaco A420 Project) that was
designed by Hafler & Keroes in 1951. It is a justifiably famous circuit,
for the use ultralinear and a nicely balanced topology.

I have some original H&K articles in my paper files about
using CFB as the conveyor of NFB to the screen circuit.
Its also gets applied to the g1 circuit too.
But most UL circuits just apply a fraction of anode signal to the screens via
taps on the
OPT primary, so the screens have FB applied.
100% screen FB is assumed where the screens are tied to the anodes,
and the tube then works as a triode.

(Connecting the anodes to the screen taps, and screens to the anode
ends would give you over 100% screen FB, reduce Ra to lower than triode,
but not exactly practical for technical reasons I have no time for now.)



Apparently, V1 is a differential phase splitter with gain? I don't
understand the operation of V1b . . . Why would the grid be connected to the
junction of R6 & R7? It looks to me like the output of either V1a or V1b
would appear at the grid of V1b, and why doesn't it oscillate?

If you send me the circuit, since I don't get ABSE properly,
I could perhaps advise.



I like the concept of the mu follower, and amps that I have built using it
sound better than ones that don't. I could see how one could be used with
V1a, but I'm totally clueless about V1b. If a mu follower could be
employed, might a 6SN7 be used for V1 instead of the higher mu 6SL7?

Maybe I should just leave well enough alone and implement the original
circuit . . . .

Sorry to return once again to the subject of tubes, but I think this circuit
is interesting and would love to know more about how it works.

Tia,

Jon

(a.k.a. "the dishonest garage trader, to which Andre shall add, the
"ignorant" dishonest garage trader . . . . ;-)

Never mind the noise, its impossible to go through life without
experiencing it...

Patrick Turner.

"Jon Yaeger" wrote

Apparently, V1 is a differential phase splitter with gain? I
don't
understand the operation of V1b . . . Why would the grid be
connected to the
junction of R6 & R7? It looks to me like the output of either V1a
or V1b
would appear at the grid of V1b, and why doesn't it oscillate?

I like the concept of the mu follower, and amps that I have built
using it
sound better than ones that don't. I could see how one could be
used with
V1a, but I'm totally clueless about V1b. If a mu follower could
be
employed, might a 6SN7 be used for V1 instead of the higher mu
6SL7?

Paraphase. Hoped someone else would answer before now. Loads of refs
in our archives. Fred liked them. Probably still does. Quad is a
famous example.

It includes negative feedback from the point of view of V1b, to
ensure in principle that the sum of the two signals to the power
stage equals zero. It doesn't oscillate because you design it
carefully.

Mu followers can be used in a paraphase, but you need considerable
reworking to do it. Considering several features of the paraphase
may reveal why.

Firstly, distortion produced by V1a is reworked and supplemented by
V1b, so the two phases are never quite the same shape. Secondly,
there is delay between the two so they may not always be exactly in
anti-phase. Thirdly, the feedback used to partially correct these
problems reduces the source impedance of the inverter so they don't
match in this respect either, so the two phases will have different
bandwidths.

Two kinds of local feedback may be used. One is from the sum of the
output voltages, and is voltage-applied to the inverter's grid, and
one is from the sum of the cathode currents by means of a shared
cathode resistor (making a short-tailed pair). Yours may have both,
but certainly has the former.

All of these features are played off against each other to arrive at
the design compromise. It is a fairly serious design problem to get
this right with mu-followers. Considering the inverter already has a
lowish output impedance and lowish distortion on account of its
feedback, you may wish to convert just V1a.

Keep in mind that a mu-follower is constant-current, and so isn't
degenerated by an unbypassed cathode resistor on the bottom. So you
couldn't share a cathode resistor to help maintain balance. The
global feedback loop would need reworking too.

There was on the net a mu-stage paraphase somewhere. Attributed to
Menno van der Veen if I remember right. Searched but can't find.

cheers, Ian