Re: No Interconnect is the Best, attenuation and source switch options.



Andre Jute wrote:

On Jan 20, 4:50 am, Patrick Turner <i...@xxxxxxxxxxxxxxxxxx> wrote:

Cost effectiveness like class A efficiency percentages of triode SE amps
are not
sonically obvious except that both lead to very good sound
providing the circuit design topology and use is optimized and sensible.

Patrick Turner.

The thing is, people like us *start* design and construction far, far
out into the ever-decreasing marginal utility fractions where an
improvement of a part of one per cent can easily double or triple or
quadruple the price of a component. You add little bits of quality to
an amp, carefully selected and matched tubes, lower ESR caps, lower
excess noise resistors, the correct weight of wire, shielding where
required, good solder and clean working, a stepped attenuator rather
than a scratchy and inaccurate pot, but you can't precisely say that
any individual piece sounds better or why. However, all the
improvements are cumulative (in fact, a better pot is amplfied
throughout the amp...) and eventually all those tiny decisions to
choose quality over crap together become audible in the gestalt and
then you know it was worth it.

Well yes you could always say the amp with an exotic and perhaps
expensive price list
will improove sound in a large number of tiny increments adding to a
good solid marginal
improvement.

A scratchy pot is not tolerated here for very long, and I do prefer
resistors which are linear with temperature and applied voltages.

But what I said was that even with the very nicest selection of parts,
and amp not optimally set up cannot be as good as one that has been set
up better.
Load mismatches are the very worst evil in so many amps I have had to
try to re-engineer,
and good bits don't address this very common problem.
ARC and CJ and many others commonly select loadings for very high power
max, low % of pure class A,
and basically they all now make pretty looking PA amps with mediuum NFB,
and medium measurements,
and sound that although is initially impressive, can easily be improved
upon, but
IF they only were to sacrifice some maximum power.
But no, marketing cowboys who now guide the design process
scream "Watts, more Watts!!!!" at the apprentice engineers.

When we come to SET amps, appalling loading mistakes are very common in
both output stages AND driver stages.
The apprentices sometimes have appalling basic knowledge of anything!

The 845 amp I have just got running has exemplary measured performance
for an SET amp. I gave the OPT two ways of arrangeing the secondaries in
a no-waste
constant current density manner with a section of either 4 ohms or 6.6
ohms.

The usual load for all negatively biased class TRIODE A1 SE output
stages is
(Ea dc / Ia dc) - ( 2 x Ra ) where Ra is at the Ea/Ia quiescent point.
For beam tetrode its 0.9 Ea / Ia, and a slightly different set of rules
apply, but lets just stick to triodes for
simplicity.
Thus you'll get a maximal PO allowable with cut off and grid current
occuring at equal
extremes of Ia. Max THD for Triode with such a load in the 5% area
usually, so if the PO max is high, then
at a few watts likey to be used, THD is pretty low.
THD increases dramatically if the load is reduced to 1/2 the above
calculated figure.
But I see samples where the makers have set it all to make max PO into 8
ohms, but
most modern speakers are anything but 8 ohms, 6 ohms average if you are
lucky and
often with mid band dips at say 300Hz to 3 ohms.
No amount of fancy shmancy R&C etc will make the amp perform better than
it does with such an appalling load mismatch.

So if any error in loading is ever allowable, it is to make the load
HIGHER in ohms than too low.
This ALWAYS means the OPT must have more turns or iron or both and many
makers baulk at
giving their lamentable products such wonderful attributes as having a
flexible OPT
which is very generous in size, weight, bandwidth and with low losses.

In the case of my 845, it is impossible to escape from the fact the
output stage
will make 5% into 4 ohms at 50 watts ( 2 x 845 ) when loaded with the
load of 4 ohms for
approximately max PO according to the above formula.
Since learning that my customer is using quite good quality "4" ohm
large floor stander speakers by Piega,
I selected the 4 ohm tapping, reduced Ea a little, increased Ia for max
PO into 3 ohms,
while still allowing considerable voltage headroom for bass frequencies
where there
will probably be a high resonant peak.
With the amp set up like this the spread of power, amount of power is
good, and distribution
of current and voltage abilities is about optimized.

The EL84 driver stage has to make a maximum of about 120Vrms to power
the 845.
The 2H distortion gives maximal 2H cancelation at 9 ohms load with
KRAudio 845, and with Chinese 845
it was at 6 ohms. Cancelations are such that at 9 ohms at low levels
there is ZERO 2H,
and only a small amount of 3H and higher H.
I tried lowering the resistance loading on the EL84 drive stage
to benignly increase the driver 2H and therefore produce maximal 2H
cancelation
at a lower load where the output stage has more 2H.
But I didn't like the spectra change.
So I have left the driver stage alone with its approx 2% THD at 120Vrms
drive.
The means by which one IS ABLE to effectively reduce THD in most SET
amps
WITHOUT increasing NFB, or indeed using any NFB is to raise the anode
load,
so it could be more like Ea dc / Ia dc, simple, but then the max PO
will may become only 30 watts instead of 50 in the 2 x 845 case.
The drop in THD is quite beneficial if no stage in the amp is
deliberatly
set up to distort to give cancelations where the output load is low.
Just about any SET amp will contain the same cancelation phenomena
occuring at some value of RL.
All will benefit from maximizing 2H cancelations over as wide a range of
loads as possible.
Because SE triode output stages produce less THD as the load value is
raised, and we usually like to have linear driver stages
with high R values then for low total THD a higher RL is needed in the
output stage.

SO, how does one overcome the problem of matching when its not available
as a
result of maker using say only ONE lousy winding set for 8 ohms?
A matching transformer from ZeroImpedance is a real boon.
These toroidal auto trannies allow a 4 ohm load to appear to the amp to
be somewhere
between say 6 and 20 ohms.
If set up for 20 ohms, the amp becomes less sensitive, and cannot seem
to make the full
amount of spec'd power, but if 25 watts is enough rather than 50 watts
max possible,
and THD is 0.02% at 3 watts instead of 0.2%, then something major has
been achieved.
Also the step down tranny likie this improves the damping factor.

I don't have shares in ZeroImpedance, but two guys here have bought them
to find the sound
became better. One guy got the trannies in their cheap form and
un-boxed, and I boxed them for him
in screwed plywood full of compacted sand and with 4mm banana sockets.
I tested the trannies. at 30 watts BW was from less than 10Hz to 1MHz,
and losses were negligible, and capacitance was low.

Replacing generic Wima polypropylene caps with Auri caps didn't seem to
make any difference to my ears.
And you'd NEVER see any measured betterment. But changeing load matching
ratios
with a load matching tranny makes a huge measured difference to THD and
possibly also to the sound.
Some audiophiles will cling to the idea that transformers all veil the
sound and that
more of them always places more veils. That's simplistic bollocks.
Veils simply depend on the physical specs for the trannies.

So the moral of this story is that any SET amp can give better an the
usual mediocre performance
achieved usually by means of intelligent loading.

So instad of 3.5k load for a 300B, it will sound better with 7k.
Sure it will then clip asymetrically, but well away from clipping
the Rout will be 1/2, the THD about 1/3 or much more depending on the
amount
of natural and unavoidable 2H cancelling between driver and output
tubes,
and sound will be better.
If there is not enough power, use MORE 300Bs, and parallel them.

The cost per watt can be enormous for an decently performing SET amp.
So what? Why do people worry about such costs when they spend absurd
amounts
on cars, house extensions and pleasing silly women ?????

If you like music, then spend what you need to get where you want.
Don't complain here about it when you have just been forced by she who
must be obeyed
says she wants new carpets and and a better car.

If people were more consistent in their money spending habits
they'd seek out the deeper truth about what makes good sound.

And maybe that new carpet might be better than the timber floor in the
listening room....

Better than spending zillions on silly parts.....

Patrick Turner.







Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review
.

Relevant Pages

  • Re: Most Linear Region
    ... Do you have any link to e810F triode curves, or did you plot your own? ... E810F triode tests with 8k load, same Ea Ia as above, gain is stated at ... Ra has to be very much higher than the 940 ohms I figured out ... Here is an amp recipe anyone may make, and find it simple, and find it ...
    (rec.audio.tubes)
  • Re: S-30 Mk III circuit help
    ... The reason is that the load match for the tubes is so very poor. ... so 5 in parallel amounts to 56 ohms. ... the amp will overheat the output tubes. ... simple combination of non linear devices and enormous amounts of NFB, ...
    (rec.audio.tubes)
  • 845 tests for thd, power.
    ... I continued to de-bug the 845 amp I have built, ... Distortion, THD, with 9dB global NFB with 6 ohm load. ... THD increases with loads below 5 ohms at all levels, ...
    (rec.audio.tubes)
  • Re: valve amp -> speaker impedance mismatch
    ... This is a greater resistance (not ... The higher the ohms of the load, the less current change in the load you ... The more bricks, the harder the truck works, but with an amp, the more load ...
    (rec.audio.tubes)
  • Re: Heads, cabs and wattage
    ... This cabinet can only have 4 each 16 ohm drivers. ... volume this would definetly change the sound of the loudspeaker. ... I don't see how the Vjr analogies are effected by what my other amp is? ... at 8 ohms, I run the amp in stereo mode, but with mono signal anyway. ...
    (alt.guitar.bass)
Loading