[AMRadio] push pull output XFMRs and neg feedback

Jim Candela jcandela at prodigy.net
Sat May 30 22:07:09 EDT 2009

Here is part of a good thread on the rec.audio.tubes list:

Andre Jute wrote:
> On May 28, 4:20 am, Jon Yaeger <jon... at bellsouth.net> wrote:
> > My recent post about the phasing of OPTs came up because I'm configuring 
> > a
> > tube amp based around the iron from a Sansui 1000A, -- the OPTs are
> > reputedly made by Tamura or some other high-quality winder.
> >
> > The OPTs are not UL, which I'd prefer.  I could regulate the screen with
> > Patrick's suggestion using series zener diodes, but I'm a little 
> > paranoid
> > about zener noise.  Is it audible?  Or visible on a scope?
> >
> > Anyway, I was looking at a 19 page Mullard spec sheet and noticed that 
> > they
> > show a pair of EL34s in triode mode in PP at 14 watts/ch (enough for 
> > this
> > application) but with just 0.4% distortion using an un-bypassed cathode 
> > bias
> > scheme.
> >
> > Bypassed the same scheme produced 19 watts but at 1.8% distortion - more
> > than a 4-fold increase.  Doesn't seem that the 5 extra watts are worth 
> > it.
> >
> > The Ra-a goes from 5k for bypassed to 10k for non.  My tranny's nominal
> > impedance is 6.6k.  I don't expect too much of a problem.
> >
> > Has anyone built a comparable EL34 amp in triode mode and left out the
> > bypass caps?  Would be interested in any thoughts or observations.
> >
> > Thanks.
> >
> > Jon
> You're not comparing like with like, feller. Caps back when that
> booklet was written were horrendous. You can do a lot better than 1.8%
> with modern caps.
> That said, leaving off the bypasses will increase the perceived bass
> extension.
> But even with that also said, I can't quite find bring myself to
> suggest you forego all the advantages of automatic adjustment of your
> amp that the cathode bypass caps provide. I like bypass caps, or in
> their place batteries.

But batteries merely offer fixed bias which isn't adjustable and tubes
do change over time.

The bypass caps don't do anything to regulate the dc bias; it is the
resistance which regulates the dc bias, with or without a bypass cap.
But the typical Rk of an EL34 PP amp would be say 470 ohms and this has
whatever plate signal current in it plus some screen dc snd screen ac.
In true pentode mode and in class A at typical op conditions the Ra is
increased from about 15k to 65k with an unbypassed 470 Rk.
In triode mode the Ra will rise from about 1.3k with bypassing, to 5.5k
with no bypass cap, or similar to what you'd get with a UL stage with
30% taps. At least 15dB NFB is needed to reduce the Rout to get a
reasonable DF of over 5.

In triode mode, the unbypassed Rk wastes some precious power.
if there is an advatantage, it is that you have local current FB, but it
amounts only to maybe 3dB, which is a negligible amount of NFB, and thus
not worth crowing about.

There is another way to avoid the waste of PO and to get the class A OP
triode stage to keep the dc balance with individual Rk, yet not actuall
bypass both Rk conventionally.

And that is to use the bypass caps, sure, say 1,000 uF each, but have
both their negative ends taken to a 10k resistance to 0V.
The 10k is low enough R to bias the caps properly, but effectively you
have 500uF cathode to cathode so they are effectively tied together at
ac but not at dc. If each Rk = 470 ohms, the effectively you have a
common Rk = 235 ohms.

Williamson only used one Rk, and it allows better cancelling of 2H
currents of each tube while ever the amp works in class A which his amp
was designed for.

The one Rk gives attrocious regulation of Ia for each tube though so I
never do it.

Once the amp works in class AB, THD rises alarmingly with a single Rk
for both tubes unless you have some means of keeping Ek constant which
is why fixed bias is chosen for AB op. A PP triode amp with EL34 can
give about 1.5% THD at just below clip if the driver amp has negligible
THD; they don't, so always expect 2% minimum THD from PP triodes without
GNFB, with pure class A1.

In class AB where the AB PO = 2 x class A PO the THD may be 4%, and with
pentodes perhaps 5%, and its all odd order junk, mainly 3H, some 5H and
7H etc.

Leak used normal cathode bias with 2 x 600 ohms plus 2 x 47uF. OK, but
that 47uF was a lousy value, and should have been 470uF.
Lousy deals were wot yer got from tube amp makers in 1955. Quad-II had
180 ohms common Rk plus 40uF. Total bean counter bull shit!
Williamson had common Rk and no bypass caps but had adjust pots. I've
seen many that went up in smoke, and again, it was attrocious
engineering. There were no LEDs and transistors to build a nice little
bias current monitoring circuit so you would know when a tube was
becoming unbiased, and you would be protected against an errant tube
losing bias control and toasting 1/2 an OPT winding. No nice little SCRs
which react to over bias, and turn the amp off for you when you have
forgotten to turn it off and a tube goes west at 2am.

Meanwhile, Allen Wright thinks that having a common CCS to commoned
cathodes of the two OP tubes is the best thing for PP **pure class A**
OP stages. This effectively has the PP tubes working "seamlessly" in
series on the RLa-a and makes the OP stage drivable at only 1 grid with
the other tied to 0V, so a single ended drive stage is used without any
"phase splitter". The drive voltage then becomes rather high if large
tubes are used.
But it works a treat with EL84, but to get some decent triode PO you
need a lot of them. With added complications, you can include separate
Rk to all tubes used with conventional bypass caps but where *all* these
things are normally taken to 0V, you have a CCS down to a negative
voltage rail. Or even a choke!!

With a high Z in the common cathode circuit there can be virtually zero
common mode current signal in the two OP tubes. Therefore the absence of
such currents leads to less IMD of all the wanted differential currents
which comprise the wanted audio output currents.

Such a PP circuit is no worse than an SE OP, but gives twice the PO of
one tube but with much less THD. You cannot have class AB PO which is
the situation with an SE stage. Everyone should realise there are plenty
of SE amps which sound better than PP amps if they have enough watts,
and all the watts are pure class A. Many PP amps are forced to work into
heavy AB to make the same PO and the sonics are worse than the SE amp.
I'd rather have two EL34 in parallel SE triode and making 12W than have
a pair of EL84 in pentode making the same 12W in low bias PP AB1.

Patrick Turner.

More information about the AMRadio mailing list

This page last updated 24 Oct 2017.