I finally have hooked up the stereo amp which uses 2A3
to my speakers which are about 90dB sensitivity.
I find I could listen all day and night to this underpowered
amplifier.
Whatever its doing it is doing it RIGHT.

The conditions are as follows.

Power tranny is Hammond type 370FX, and with
a 240V: 2 x 2.5V filament tranny.
Hammond 10H choke used with a quad of Sonic Frontier
PS caps of 1,100 uF each, so that with seriesed caps
I have 550uF, 10H, 550uF, as the CLC to supply power to the
pair of amp channels.
Rectifiers are IN5408 SS diodes. I don't strictly believe in tube
rectifiers;
I like them, I even use them sometimes; I just don't think they are
indispensible.
But in this case the use of a tube rectifier would have reduced the
available B+
and forced me to use fixed bias, which I don't like for any SE amps.

No need to dual mono power supplies.
180uF used to further filter the input and driver stages.


Output tubes are 2A3, RCA, NOS?, maybe used, not sure.
Ea = +270V, Ia = 57mA, cathode bias with 820ohms
bypassed with good quality 100V x 470uF.
heaters are AC 2.5V with a pot for hum nulling.

OPT is a 2.5k : 4/8/16 Hammond 25 watt rated SET gapped OPT
type 1627SE.

OPT is set up to match 2.5k to 4ohms, but I find 5 ohms is the most
correct load.

At 4Watts I get excellent open loop BW with this OPT
and freedom from saturation, of which there is only a slight trace at
about
7 Hz.

The OPT has queer resonances above 30kHz, makinf the use of
global NFB difficult for ppl with no clue about how to overcome such
connundrums.
It must be due to primary being wound close to the secondary,
and the absence of enough interleavings of windings, and the absence of
sufficient
insulation thickness.
Nevertheless, the 1627SE did turn out to be very usable, and sound is
glorious.

Driver tube is an ancient Radiotron 'made in australia' 6SN7,
used but OK. 1k Rk bypassed with 470uF, RL = 34k, 0.47 uF Wima coupling
cap,
150k grid bias R for 2A3, 1.8k series grid stopper.

The 6SN7 is operating fairly linearly and comfortably.
Without Ck bypass, the driver thd reduces, but then overall thd rises
a lot since less even order Dn cancelation takes place.

The fully bypassed Rk on the 6SN7 gives lowish Ra of about 5k for the
paralleled tube
and is thus a sufficiently low enough drive resistance to enhance the
micro detail behaviour of the
2A3, ( very necessary imho to get the best out of not just a triode
output tube,
but ALL output tubes.)

The input tube is 1/2 a 12AU7 of unknown lineage; it measured nicely,
and I like the 12AU7 for inputs; its warm, accurate, and sounds
glorious.

Its Rk is 1.8k, and also bypassed with 470uF, and taken then to 100 ohms
for
the global NFB to be applied.
RL is 50k, 0.47uF to a network for LF gain stepping, 1M+0.033uF
plus 220k, typical in many of my amps to vastly
improve the bass stability when the 12dB of global FB is applied.

The FB resistor divider network is 750 ohms to 100 ohms
with a phase tweaker bypass cap across the 750 ohms = 1,000 pF.
There is a 470pF plus 2.7k zobel network connected from the
driver grid to 0V to control the HF gain and ultimate phase shift of the

12AU7 to 6SN7 interface.
A further 5.6ohms+0.47uF Zobel network is across the OPT secondary,
and finally I got the amp to be fairly ring free when connected to
any value of cap without a pure R load, and when using a square wave.
Open loop BW with all compensation zobels in place was 20Hz to 20kHz,
-3dB,
and with 12 dB of FB the full power BW is 5Hz to 55kHz, 5 ohms.
It remains the same at lower levels of power.

A 5 kHz square wave is a bit ragged due to the mentioned OPT design
simplicity,
but the sine wave response between 30Hz and 20hHz right up to clipping
is dead flat.

A 50k Alps Black pot is used for the gain control.

Input sensitivity with 12dB applied global NFB is 0.7Vrms.
The amp's use with an CD player is fine.
Just no need for any preamp.

The sound is that of a far more powerful amplifier,
with tremendous sense of rightness, instrument detail, warmth without
muddle, not the slightest hint of any lack of bass, the usual beautiful
mids, and detailed top end.

The amp measures well at less than 1% thd at 4 watts into 5 ohms,
mainly all 2H.
But at normal levels of a watt the 3H and other H are present below the
2H
which can be seen on the CRO when monitoring the THD at all levels.
Ro is 0.47 ohm, and although my speakers vary between
4 and 20ohms, the sound appears firm, well timbred, and there isn't any
hint
of the sound being that from an inadequate amplifier.

I played Handel's water music, the sound track of 'The Pianist' for my
tests plus
a not so wonderful Naxos CD, 'The Best of Bach'.

Who says SET amps suck?

Certainly not I!.

Methinks that if the owner wanted more power, he could simply swap the
2A3 for a KT90 since the PT has a 6.3V heater winding, and then
remove the series R I have placed in the B+ lines to trim the B+
to where I want it, and with 270V, and with the UL tap on the
OPT that is provided, maybe 9+ watts of SEUL would be available.

The now unused 1670 OPT UL tap is at 40%.
But it could be at 60%, if you use the plate winding reversed,
thus further drifting a tetrode towards triode, but still not
suffering a huge loss of power due to grid current limitations.

300B could also be used, and with slightly more Ia,
but with the same low Ea, thus suiting moderate power and good matching
to
low value loads, which is a major failing in many SET amps;
they try to match so max power is at 8ohms, but that is
nearly always wrong with modern speakers, and
max power should occur at 3 or 4 ohms.


Patrick Turner.