Re: pentode amplifiers



Andre Jute wrote:

On Nov 8, 3:23 pm, "Arny Krueger" <ar...@xxxxxxxxxx> wrote:
"Iain Churches" <Iai...@xxxxxxxxxxx> wrote in message

news:dw2Yi.248254$Ie3.73848@xxxxxxxxxxxxxxxxxxxxxxxxxx

"Andre Jute" <fiul...@xxxxxxxxx> wrote in message
news:1193955653.337450.202680@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Dave <dspear9...@xxxxxxxxx> wrote:

"Andre Jute" <fiul...@xxxxxxxxx> wrote in message
news:1193877935.752596.16330@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

The explanation is simple. Loop NFB causes artifacts
of ever lower magnitude but higher order.

Good example of majoring in the minors. Loop NFB drops nonlinearity by 20
dB or more. The equipment in question was not entirely free of higher order
nonliner distortion, and the secondary effect being obsessed over here
typically adds far less distortion than was already there.OTOH, the loop
feedback drops all nonlinear distortion by 20 dB. The net higher order
distortion is thus dramatically reduced.

We're not talking about the net number, Krueger. We're talking about
the effect of the composition of the residual distortion.

Well yes, but with music this is an extremely complex mix
of dynamically changing frequencies, and listened to alone
separated from the undistorted signal, sounds like noise and range of
harmonically and unharmonically related tones; bloody awful sounding,
worse than pink noise or any real music.

Its the aim of every tube amp maker to reduce this to below audible
levels
without applying too many tricks.



These high
order distortions, even at 60dB below conscious
perception are very, very disturbing,

If they are so disturbing, why aren't the SET owners
running out of the room
screaming every time they turn their MI amps on? Repeat, its not like SET
amps are free of higher order distoriton.They are based on tubes and tubes
are exponentially-based devices. The expansion of their theoretical
amplitude transfer function includes signficant higher-order terms, loop
feedback or not!

A conservatively designed and well-developed 300B amp can easily get
the third and higher harmonics down to 0.03 per cent without any loop
or stage feedback. I don't see how that is "significant" at all.

Hmm, most SET amps with 300B need to be powering horns so THD/IMD
is less than 0.2 watts and then you might get down to 0.03%.

But with ordinary speakers of 88dB/W/M, the SE amp with one little 300B
will have at least 3% at 5 watts with a highish RL, and at 0.5W average
power for such insentive speakers
the output voltage is about 1/3 that for 5 watts, and THD is 0.75%,
mainly all 2H, but
way above 0.03%, which is only realizable if you have lots more 300B in
parallel
or you have a PP pair, and powered by a balanced PP input stage to avoid
the 2H made in so many half baked designs
for PP where the 2H of the single lousy driver is way above the mainly
3H of the PP output stage THD.

If the output SE stage's 2H is cancelled by the driver stage's 2H then
its possible to very
much reduce 2H down to less than the 0.03% at up to a few watts, but
then that's only
for one value of load, as the amount of 2H changes with load on the
output tube.
Natural cancelling of 2H by *VOLTAGE* cancelation in SE amps instead of
*CURRENT* cancelations
in PP stages actually makes tube sound better, so my clients tell me.

In a generic SET amp with 300B, it is thus possible to make a triode
driver stage
which makes 0.75% of 2H at the 0.75W output level, but its also not
wise,
since at that low output level most well designed driver stages won't
make that much 2H to cancel anything much. A driver stage naking such a
lot of 2H will struggle
at higher levels, and maybe clip earlier than the output stage.
IMD produced in the driver get passed on and muddy the output spectra.

Hence my solution is to use local CFB from the OPT around
tetrodes/pentodes, and reduce max
output stage THD to say 2% at 20 watts, abd then the driver stage when
loaded with an RL < 10Ra
will never struggle or muddy up the sound, and what little 2H it makes
willo
nicely cancel the output stage 2H....and presumably most other even
order products.
The result is beautiful un-muddied SE sound, and because the amps I make
have
over 20 watt capability, there is enough headroom for most ppl
and perhaps 1/10 of the THD/IMD as one may hear with a lone 300B with
average power at 1W.

The odd order junk isn't thus cancelled, but will be a vile residue.
But its at such a low level below what would otherwise be a large 2H
amount, it matter not,
even if you weighted the audibility acording to the Nsquared/4 rule.

So in fact in my amps I *actually do achieve* your quoted 0.03%, but
nobody else seems to.
Most SET amps I have to rewire or teach how to sing and how to behave
are travesties of engineering
because the makers fail so dismally with basic loadings on tubes, and
thd and noise is way above what is
possible.

Because all amps I make have wide bad OPTs giving over 50khz of
bandwidth WITHOUT dependance on global NFB,
when I do apply a bit of NFB, say 9dB, to ensure Rout of the amp is at
least 1/5 of RL,
and preferably 1/10 RL, then the THD and IMD is further reduced and the
net residual artifacts no matter what they
are would be much lower level and far less unpleasant if sent to the
speakers without the music signal present
compared to the rubbish and junk which exists in an average low power
SET amp signal if played a little too loud
and without distortion cancelling schemes either bt way of clever driver
or PP design, and
without loop NFB.

Audio is a gentle low level energy in the air and excites our fabulously
sensitive ears
which have evolved over millions of years.

But folks seem hell bent on owning insensitive speakers, and sometimes
will aquire low power SET amps. If they listen up close to the speakers
at low level, fine,
and I know guys who do this, but filling a large hall sized lounge with
high level is impossible and silly.

Its not easy to accurately reproduce the sound of a loud trombone player
right in front of you
without clipping anything. Ditto someone pounding away on a grand piano,
lid up,
facing towards you. Horns come to mind. And lotsa watts.

whereas second
harmonic up to three-quarter per cent cannot even be
distinguished by professional musicians. It is a
subliminal effect, and of course in pentode mode it is
exaggerated. People look at the total harmonic
distortion but in fact the higher harmonics must be
weighted much more heavily than the second harmonic to
account for its extraordinary subliminal effect, often
described as "edgineess" by professional musicians.

This was the explanation I was looking for. thank you.

Yes. That seems to be a good description of what is going
on, and explains why two similar tubes/valves of the same
type but by different makers may sound different. They
usually have the same or very similar THD but the
distortion spectra are sometimes surprisingly different. 2H is exactly an
octave of the fundamental, and so, in
"trace" amounts may be regarded as benign. 3H, 5H, 7H
and 9H, even at much lower levels are considerably more disturbing.

Same mythical thinking, repeated.

All right, Krueger, so you don't like psycho-acoustic truths
observable in repeatable tests disturbing your fauz certainties. So,
slumber on.

Repeat again, its not like SET amps are free of higher order distortion.They
are based on tubes and tubes are exponentially-based devices. The expansion
of their theoretical amplitude transfer function of a tubed amp includes
signficant higher-order terms, loop feedback or not!

So you keep saying, Krueger, but so far you have provided no proof of
your contention, which those of who bulld SETs know is untrue.

I have heard many excellent sounding systems with 8W SET amps with a
lone 300B.
The best had either professional horn speakers made by JBL or Altec,
or large sensitive Tannoys in huge bins for great bass.
Average power was miniscule, and issues of audible THD/IMD are not worth
discussion, because the levels are so low.

It so happens that at a watt, 2H from a 300B or other tride such as
trioded KT88, 6550, or KT90
etc is often 20 times the level of any other H.

Pentodes and beam tubes are not so benign, and produce varying amounts
of many harmonics
and 2H will be high like in a triode with too low a load when it is
loaded low,
then fall to zero at some critical medium load value, and then rise
again as RL is increased,
but the phase of the 2H is opposite below and above the critical RL.
The odd numbered H are many and varied with pentodes and beam tubes and
the
best way of dealing with them is NFB, and this is achieved in every
triode because of the
electrtostatic NFB acting between anode and space charge summing point
between cathode and grid.
The triode NFB does a pretty good job at odd order H reduction.
But if a similar amount of local CFB from an OPT is applied then the H
reduction
is similar to triode.
So to shoehorn a beam or pentode tube into low THD production,
the open loop gain must be well reduced to about the same as low µ
triodes.

So in my latest work with 13E1, the open loop gain is 16, but reduced to
2.5 only
with the local NFB from OPT CFB and THD becomes benign.

The tube sound IS THERE, but not muddied by anything very much.

The baby is retained, only the bath water is thrown out.

With BJTs, the effort to make them tolerable to listen to always
requires lotsa NFB as with pentodes or beam tubes.

Arny doesn't take in the variety and differences in tubes; triodes are
infinitely more tolerable sonically than pentodes or beam tubes or BJTs
or mosfets
when used *WITHOUT* any NFB.
That mantle radio set with a lone 6V6 is a HORRID little audio crapper
if ever there
was one, but generally has no GNFB or local NFB.
Any single BJT used similarly would even be worse..




Some years ago, the technical director of Svetlana sent
me the findings of a listening group with whom he had been working. Their
task was to evaluate 6CG7 tubes by different
makers, and put them in order of preference, so that they
could be measured and analysed. Although I was not able to obtain all
twelve makes on the list, with
some colleagues, I repeated the experiment.The differences
were most interesting. The interpretation of which is
"better" must be left to individual taste, but in general
terms, we ranked the tubes in roughly the same order as the Svetlana
listeners. In case you are wondering, the
RCA cleartop was the best sounding. It also had the lowest odd order
harmonics when measured in a mu follower circuit.

Yet another anecdote with questionable relevance.

Well here I have to agree that sonic differences in preamp tubes can be
heard.
Even when THD and IMD measures below 0.01%, a typical value where
you have a line stage preamp with volume pot before a generic gain stage
using a 6CG7.

I have, with 4 other guys spent an afternoon tube rolling with 6CG7
samples.
We agreed the NOS Seimans were best, then made in Oz Miniwatts, then
Mullard, and
then very much last were recent Sovtek versions.

SNIP,,,

Right now I'm running a solid
state amp which is rated at 50W RMS with THD of 0.08%. I guess this
isn't a
reasonable target for a tube amp unless I run multiple
parallel UL PP pairs
per channel...
A while ago, I built a push pull parallel EL34 amp, with
two parallel pairs per channel. It can achieve 0.08% THD
at full power.
http://www.kolumbus.fi/iain.churches/Pics/HomeAudioSystem/C50_002.jpg

I don't know whether to laugh or cry.

Ah well, decisions, decisions.

But the SS amp that makes 0.08% at a dB below 50W clip
could make 0.08% at 1/2 a Watt if not carefully designed because of
crossover thd.

The tube amp which makes 0.08% at 50W will have
smoothly decreasing THD as output voltage is reduced, so
that at 5W, THD will be perhaps 0.03%, and at 0.5W its 0.01%,
and utterly inaudible no matter what the harmonics are.

If the amp moves from class AB to class A at the low PO level
there is ZERO crossover H generated at low levels,
because all is handled by devices in their linear regions.

Patrick Turner.

Snip more...
.

Relevant Pages

  • Re: Best tap for NFB?
    ... it by reducing NFB. ... cuts gain by the resistor ratio. ... work) and the same increase goes into the driver tubes. ... by using the shelving network before the balanced amp, ...
    (rec.audio.tubes)
  • Re: Listening again... many feedback loops
    ... Kardon Citation II amp. ... Or is a single NFB loop from OPT to ... But there have been more arguments over whether NFB and PP reduce sound quality ... disgustingly hgh levels of open loop THD IMD and noise to low levels. ...
    (rec.audio.tubes)
  • Re: St Francis of the RATs was Re: KISS 123: Why an ultrafi tube amplifier has Zero Negative Feedbac
    ... running the SET amp where it is flat into horns. ... >> of NFB, and many commercial designs use this much. ... >> thats typical Linn topology open loop gain in SS amps. ... >>> fed back offsets not only the distortion but a goodly part ...
    (rec.audio.tubes)
  • Re: Listening again... many feedback loops
    ... Kardon Citation II amp. ... But there have been more arguments over whether NFB and PP reduce sound quality ... disgustingly hgh levels of open loop THD IMD and noise to low levels. ... more power than you should decently extract from any particular tube ...
    (rec.audio.tubes)
  • Re: ARC VT100 repairs.
    ... The set up allows balanced global NFB and balanced or non balanced inputs. ... A couple of 470 ohm R across such a winding slung between cathodes of the input LTP ... In fact far fewer devices and tubes are required. ... Like in any conventionally unbalanced or balanced amp. ...
    (rec.audio.tubes)