[Patrick Turner]

On May 21, 12:21*pm, flipper wrote:
On Fri, 20 May 2011 06:54:32 -0700, Big Bad Bob





wrote:
On 05/20/11 03:22, Patrick Turner so wittily quipped:
Baxandall's amp isn't real bad, but the first input pentode must
produce a voltage large enough to drive one 6L6 grid in beam tetrode
mode with fixed Eg2. The other input pentode operates as an inverting
anode follower with unity gain, but any THD generated in V1 is passed
on to be reproduced in V2. In effect, driver amp distortion is double
that of a single tube, and this defect also occurs in any amp using a
paraphase driver.

I hadn't heard the term 'paraphase' until you mentioned it, so I did a
bit of online research, found out that I've seen things like that before
and I never liked it done 'that way', but yeah, using PENTODES just
makes everything worse for distortion and potential imbalance. *Relying
on the circuit gain characteristics of a tube for your design, or
requiring a balance adjustment to make it work, is just silly.

Yes, well, the original Williamson called for a balance control on the
separate triode voltage amps because their gain is what it is. The
Williamson looks balanced but unless the triodes are identical, fat
chance, it isn't. (He later decided GNFB was enough to not need the
balance pot).

Williamson fell under the spell of the bean counters who had declared
war upon anyone who ever used one more resistor than necessary
acording to the rules laid down by the Society Of Bean Counters.

So he used that stupid arrangement with his KT66 cathode circuit
involving pots to balance ****. Balance it did, until a tube over
heated, and in EVERY single Williamson sample I have had to repair,
guess what - fried pots where thy just should *never ever* be used.
Leak's idea of cathode bias with two R&C bias networks was best, and
**** the rest. Its because the dual R gave the best bias regulation.

The balance of the Willy *short* tail pair, STP, was poor because the
tail R was such a low value. But the concertina as per Willy gave
excellent balance providing anode and cathode R were equal. This
balance is maintained well enough in a Willy amp, but can be be much
improved if a 4k7 is put in between 0V and the 470R which is in the
orriginal. THEN the natural balance of the STP becomes good enough to
never wish for more. But with that 4k7 added, the STP won't give very
good balance if driven on only one side, like the Leak circuit and so
many others, Mullard 520 etc. The simple answer to good balance and
simplicity is to use an LTP with CCS tail.

And if one don't like the 2H generated in V1 SE stage from ruining
music then use an input LTP with 6SN7/6CG7/6DJ8/ etc for V1 and V2,
with CCS tail, then use a second LTP for V3 and V4 with
6SN7/6CG7/12BH7/EL84/EL86/ etc, which has a tail R taken to -150Vdc.
The amp input is taken to V1 grid and GNFB is applied to V2 grid, and
the whole input-driver is thus beautifully balanced.
I recall I saw a version in RDH4, but have forgotten where, and it
don't matter because in 1955 they didn't have handy things like an
MJE340 which makes a splendid CCS.

The Baxandall is a floating paraphase, which uses negative feedback.

Look at it again. The second pentode is driven by the difference of
the two signal paths, the two 470k summing 'sense' resistors, and is
not operating open loop like the Williamson voltage amp triodes are.

True. What distracted me was that the SP61 have different anode RL for
Idc, input has 47k, and second tube has 4k7, and this looks like a
typo. Let's assume both SP61 have RLdc = 47k. But you'll find balance
only fair enough. The higher the gain of such signal pentodes, the
better the balance becomes. I don't like seeing any grid circuit
operating from a high impedance like the two 470k. Isn't the Baxandall
circuit almost identical to the Schmitt? p526 RDH4.

But ARC d40 used a circuit where the input signal came to one input
tube with cathode FB and also to a unity gain amp to make the second
phase which powered a second input tube with an oppositely phased FB
signal. The OPT had a CT and there were balanced GNFB loops of
opposite phase. Too complex of course. But the d40 was better IMHO
than later ARC which became a ****in nightmare of complexity.


I used a similar floating paraphase on my "Batman" battery guitar amp
but used triodes and since they were low gain I 'predisposed' the
balance to nominal, and let the feedback work around that, but with
pentodes you have lots of gain for NFB around the phase splitter.

The floating paraphase has a different HF roll off on the two paths
but, then, so does a concertina.

Also true, but the HF output roll off of the two phases in a
Williamson are both at such a high F that the THD thus generated is
well above 20kHz and not important. I found that the the HF balance
from a concertina could be made near equal if about 15pF or other low
C value is across the cathode R of the concertina to slightly boost
the gain from the anode.

None of them are 'perfect'.

In RDH4, P344, there is a 30W beam tetrode amp with 807 with SE
pentode input, pentode concertina.
Page 344 has the triode Williamson with 807. P353 is interesting
though, positive and negative FB loops are used. This is PURE
QUIRKINESS for those who must have quirks in their amplifier
collection. I have never seen any commercial design with PFB and NFB -
too hard to get right. But most amps do have quite a few quarks
lurking about.

Nothing is ever perfect.

Patrick Turner.