[Patrick Turner]

Tonight my customer came around for a couple of hours
to listen to the amps he's buying from me.

He brought a bundle of CDs and we started off eith the SE 845.

No listening test means much unless you have a good standard Other Amp
to compare, and in this case the OA was a VAC amp which I have
been working on for a week to teach it how to be civilised, and not spit
out
so much smoke, despite the good sound.

Today I got most of the mods to the VAC completed and I fitted a new set
of
8 x Sofia branded 300B with CARBON ANODES.

http://www.sophiaelectric.com/

For an hour we changed back and forth between my SET845 and the VAC with
new Sofia tubes fitted.

The sonic differences were not huge, with both performing very well,
but the SE had slightly more warmth, and a better feeling of 'being
there'.

Resonances in accoustic guitar and mandolin were better produced by the
SET amp.


Half way through the test I adjusted the VAC NFB switch to zero global
NFB
from the initial maximum of 12dB.

My customer preferred zero NFB; music seemed to be more musical,
despite THD/IMD rising 4 times.

Sill, THD/IMD is under 0.2% maximum for all we listened to without GNFB.
With GNFB, the max is about 0.05%.
Average is lower.

The 845 amps with 7dB of GNFB make 5dB less THD/IMD than the PP VAC with
12dB of NFB,
and with similar loading at all levels up to 2dB below clipping.
I think if I fit better B+ filtering to the PP VAC, the
difference in THD&IMD will become a lot less.
But without any GNFB, the VAC with PP 300B makes 4 times the THD&IMD
of the SET 845!
But never mind the measurements; its the sound that matters.


The 845 amps each have 2xKR Audio 845 output tubes in SET parallel,
3 x Sovtek EL84 in parallel as the driver, and 1 x 6CG7 NOS made in
Australia paralleled for the
input tube. If there is a sonic signature, it is partly due to how all
three triode tube elements interact.
To me they seem to like being together.
The speakers I have are 8 ohms to about 200Hz, then rising to 10 ohms by
1kHz,
then falling to 8 ohms above 3kHz, due to the seriesed D'Apolito
connection of midranges.
The damping factor for bass control is little better than only 3
with no FNF, but the sound stayed good and bass was not bloated.
But then when I make speakers, I make em easy to drive....

The test clearly illustrated yet again that you
DO NOT NEED GNFB with triode amps IF you know what you are doing.

And because there are 3 output terminals for 8, 4 and 2 ohms,
to halve THD/IMD and increase the DF by 2x all you have to do is move
the speaker
from the 8 ohm outlet to the 4 ohm outlet.
As RL seen by the output triodes rises, the voltage change slightly
increases while current change reduces a lot, and so does the
distortions.
OK, you cannot get as much PO using an 8 ohm speaker on an a 4 ohm
outlet,
but you only have to have enough power. Tonight we had plenty.
I could have moved the speakers from the 8 ohm outlet which is ALL
the OPT secondary winding to the 2 ohm outlet which is a centre tap at
the OPT sec. This would reduce Rout from 2.4 0hms to 0.6 ohms,
and lower THD by 1/3, all without any GNFB.

The conclusion of the comparison was that we both thought the
VAC and 845 were both extremely drinkable wines.

So, GNFB isn't always necessary with triodes.

And SET amps are NOT always worse measuring or poorer sounding than PP
amps.

Certainly not when the SE amp can make MORE PO than the maximum
available by PP.

We changed preamps to something my customer had borrowed for trial
from a local hi-fi store.
It wasn't any better than my reference preamp using 2 triodes, one for
gain and a
another for a buffer after the gain pot.

Then we tried the preamp stage of an integrated power amp
I had totally re-engineered for my customer 12 months ago.
This integrated amp is set up so its line stage preamp
using only ONE triode can run separately
with the output tubes turned off.

The preamp from the shop had a lot of complexity and gee whiz looks,
but it was dray and clinical compared to the more accurate
sound from the single lone triode.

The guy won't be buying the $2,000 preamp any time soon.

But I will have to fit a remote control....

At the end of testing, the VAC with its brand new 300B
had more hum then when we began, and it was quite audible.

At first the fitting of new tubes still resulted
in some hum exceeding 1mV, and I swapped a couple of tubes around
which reduced the hum to less than 1mV, but
with the least best tubes not exactly equal in Ia or gm on each side
of the PP circuit, hum can be worse than an SE amp.

And the tubes tend to 'bed in a bit' after some initial use.
Despite matching by the maker, they change with use.
PP amps are more affected than parallel SE because in PP you have to
*balance things*, in SE, you don't have to.
The VAC amp has no means of adjusting hum to a minimum.
There would beed to be 8 adjustments perhaps, one for each 300b,
and this is impractial.
Better would be to further filter the B+ so that the very slight
Ia difference in each side of the PP circuit
will not cause a difference in Ra and gm and hence a differetial
signal of hum across the OPT which then gives you hum at the secondary.

So to complete my upgrade of the VAC, I MUST fit a second filter section
to the anode supply; probably it will be a couple of 470uF in series and
a 120 ohm resistance.
Not enough room under the chassis for chokes, and besides, the
R&C isn't resonant, although there is 9 watts of heat in the 120 ohms.
But the hum at the CT will be reduced from 1.8Vrms to 0.1Vrms, some
25dB,
so hum at the output will greatly be reduced even if Ia is a little
unbalanced.
IMD caused by too much hum in the anode supply will also be reduced.
But there is still 66mV of hum across the cathode from the Vdc appied to
each cathode.
I am hoping this won't matter once the B+ is properly filtered,
so before I add another stage of filtering, I can temporarily
connect a few hundred uF across the existing caps to see if hum reduces
at all.
If it doesn't work, its back to the drawing board.

The VAC has 4 little power trannies supply the ac power to diode bridges
and 33,000uF at
each cathode supply. These trannies take up a lot more room than a
single 50VA tranny
would with 8 windings for the 300B cathodes.
I may have to romve the 4 little trannies and replace them with one 50VA
type,
and then have room for CRC filtering for the 8 x 5Vdc heaters each
needing 1.2Adc.
If each C = 33,000uF, and R = 1.5 ohms, the ripple could be reduced from
66mV to 2mV, and there is no need to have two 27 ohm balancing resistors
to each cathode.


When you know an amp is going to be used without global NFB, every
possible avenue
must be explored to use passive methods to reduce noise and distortions.

My customer has a better CD player than my old Denon, and perhaps has
better
speakers, Piega floor standers of some merit, so things can only
get better for him in future when I finalize the metalwork covers and
grilles
over the new 845 amps.

Patrick Turner.