[Ian Iveson[_2_]]

On Dec 8, 10:31*am, Patrick Turner wrote:
Ian scribed........

SRPP and mu-follower are essentially the same circuit, but the mu-
follower has a larger current-sense resistance for the current source,
with a tap for the bias, which then calls for AC coupling from bottom
anode to top grid.

Well, SRPP and mu-foll *do have "totem pole" config but that's where
the similarity ends.

The difference is as I have described it. The difference in how you
envisage the operation of the two circuits is up to you.

Alex has already pointed out that the gain of a symmetrical SRPP must
be mu/2 and I have agreed with him, giving supporting analysis. I
provided a link to Kimmel's original paper which develops the mu-
follower from the SRPP.

Do keep up.

The SRPP can only be really called PP when there
is substantial work on the load done by BOTH top AND bottom tube, and
the load is such that the top Rk generates enough load current from
bottom tube to give Vgk applied to top tube which is nearly equal to
the Vgk applied to the bottom tube from a signal source.
Suppose one had a pair of EL86 in triode ( similar to EL84 in triode,
but Ra = 1k2 ) in SRPP circuit, and for each tube Ea = 150V, Ia =
60mAdc, and phones of 32 ohms were connected to top cathode, then one
probably could get about +/- 50mA peak Ia change in load to give 39mW
of power into 32 ohms which to the tubes looks like a short circuit,
because an ideal class A load would be 1k2, with each tube "seeing"
close to 2k4, or R x Ra.
EL86 has u = 10 approx, and gain with EL86 is low with load of about
64 ohms, and THD is high, so you'd need a shirt&trouser load of NFB to
straighten out the mess you have without an OPT.

It would be easier for you to simply link to Kimmel's paper. Except
it's already been done.

But suppose you had a mu-foll, with say 6AS7/6080 with Ea = 100V and
Ia 200mA, with choke between cathode and 0V. The cathode voltage would
have a cap to keep Idc from the phones, say 2,200uF.
The driver tube can be set up like a normal common cathode gain stage
except that its dc RL is a pair of resistors to B+ with junction
bypassed to 6AS7 cathode with say 470uF, ie, top of choke, and also
the signal output for the phones. This means the output voltage from
"top" tube 6AS7 is in series with bottom tube anode supply resistance,
and this resistance is barely aware there is any load connected; and
in fact the load resistance is "bootstrapped" and appears to the drive
tube as a much higher R than it actually is, so the driver tube
provides low THD drive to the OP tube. With 200mA in the 6AS7 at idle,
load current can be maybe +/- 150mA giving 0.35 Watts which is enough
to deafen our Ian if the Rolling Stones is played that loud.

So the mu-follower has very little load current produced in its bottom
tube and this is also true in a preamp situtation such as athttp://www.turneraudio.com.au/Line-preamp-2003.htm

The use of mu-foll as opposed to SRPP is to AVOID load current sensing
between top and bottom tubes and to promote the top tube to being an
almost pure cathode follower,

Had you read my original post on this thread, you would realise that
the the circuit under discussion does exactly that.

and then have the bottom tube operating
with a very high effective RL which its THD becomes much lower than
SRPP, or a normal tube with non bootstrapped resistance for dc to
anode.

I'm quite looking forward to experimenting with my
emerging headphone amp, but I'll likely end up with just a single
generous triode and a beautiful output transformer.

In the last headhone amp I used EL84 in triode with OPTs taken from a
Fisher AM/FM receiver.
The load was about 5k : 16, and for headphones, ANY load is about OK,
you'll have enough power/voltage headroom.

Yes. In my case 5k : 62R

I'm still dithering on the sec. 4 x 4R or 1 x 62. If only I could
predict the effect on leakage.

I used 1/2 6CG7 to drive EL84 with 12dB NFB.

With 6CH6 in triode I would have more or less unity gain, and enough
headroom. It would be nice to try a proper triode though. If I can do
without feedback, that's what I'll stick with.

The customer is extremely happy.

Lots of headphone amps are rubbish. Not many ppl used to care because
even poor headphones can sound much clearer than a poor room system.
There's a huge market for headphones now, although much of the top end
is functionalist rather than functional. Within that market, a new
generation of discerning audiophiles is emerging.

Noise is the main problem to overcome, so I suggest all trannys are
potted, and DC is applied to all heaters; an external PSU is a good
idea. Even then, tube noise can be present. One should aim for noise
to be less than 0.05mV, or 50uV. You may find this impossible achieve,
but you get say 0.25mV. Well, a simple resistance divider across the
16 ohm winding, say 39 ohms plus 8.2 ohms will give mean the tube
signal level will have to be about 6 times higher, but the R divider
divides the noise down so SNR is improved by 6 times. Most headphone
outlets on power amps and receivers have such a divider which allows
tubes or transistors to see a high value load and hence give low THD,
while being able to still give oodles of voltage for headphones.

Noise is indeed the biggest...actually the only...problem to be
overcome.

I'll start with as much as possible and then reluctantly add bits
until the noise is gone. That way I'll know what I can't hear.

The phones amp I built for my customer in 2009 is fully integrated,
has hi & low level outputs, and can be used as a normal preamp with
superb performance.

A lot of ppl try to avoid OPTs in phone amps while insisting on tubes
to give OTL type of power. Its not good practice, but a pair of
complementary source follower mosfets could easily be used with a +/-
12V supply
and Ia at 0.5 amps, and then one drives both gates from a C&R coupled
triode of some sort. The secret is to have the voltage amping done
linearly, and also the current amping, and class A mosfets in source-
foll mode will be remarkably linear, even if the phones are 4 ohms.
6AS7 with a PP OPT would also be OK.
But once you go to tubes and OPT, the OPT allows a normal amp to be
made, and loading and linearity becomes much better than without OPT.

That's why I'm using an OPT. Otherwise, Nat Semi does a very nice
headphone buffer chip and I wouldn't need valves at all.

Ian