[Patrick Turner]

Trevor Wilson wrote:

"Iain Churches" wrote in message
i.fi...


"Trevor Wilson" wrote in message
...

"Iain Churches" wrote in message
i.fi...


"Trevor Wilson" wrote in message
...


A SET amplifier rated at (say) 10 Watts @ 8 Ohms, can only deliver a
maximum of 5 Watts, when the impedance falls to 4 Ohms, 2.5 Watts @ 2
Ohms
and so on.


Trevor, you are going round in circles. Please study
IEC/EN/BS EN 60268-5 to which most
competent loudspeakers confirm.
The exception being ESL.

**ESLs ARE competent loudspeakers.

No one suggested they were not. They are the
common exception to IEC/EN/BS EN 60268-5.

When asked about this, Russ Walker smiled and
said that he belived his father Peter had "received
a dispensation from the Pope"

That satisfied everyone:-)

**Except SET owners, of course. SET amps will have problems in coping with
what is widely acknowledged as the most accurate form of music reproduction
system. Seems like a bad trade-off to me.

More BS from TW.

If the SE Triode amp, or SE Transistor amp, or SE Mosfet amp or SE Beam
Tetrode/Pentode
amp has been designed to cope with a given ESL, it will cope just as
well as any other amplifier!!!!

Trevor omits to inform us of the full truth.

Now most full range ESL such as the Quad ESL63 and some older models by
Martin Logan
present to an amplifier a supposedly horrid sort of load varying
from high Z at LF, ie, very easy to drive at LF, to a low Z of maybe 1
ohm at 20kHz, or even less,
and perhaps rather like a C + small value R in series.

Not only is the load vary variable in Z, the ESL tends to be
insensitive, requiring
a large voltage ability at LF and a large current ability at HF.
But overall, most ESL require more power ability than dynamic speakers
merely
because there has to be both high voltage at LF and ENOUGH current
ability at HF.

But fortunately, most of the audio energy in music is between 100Hz and
1KhZ,
and anyone who has listened solely to the audio being fed to a tweeter
with Xover point at 3kHz
will be surprised to find how little music energy exists between 3kHz
and 20kHz.

What we do not want is the amplitude of this small amount of power
signal modulated
by LF wave production, whose amplitude is usually many times the HF
voltage amplitude.

So speakers like ESL57 are comfortably driven by the amplifiers such as
the Quad-II,
a humble amp if ever there was one, compared to today's behemoths with
10 times the power ability.
But Quad-II struggle with ESL63 and later Quad ESL.

Quad-II produce 20 watts into 16 ohms class AB, and
the reducing ESL load down to 1.8 ohms at 18kHz will stifle any attempt
the amp makes to produce 20 watts at 18kHz.
But the amp will never ever be needed to make such a huge amount of
power at HF,
unless the owner is mentally unbalanced and likes ear crushing
recordings
of ppl bashing cymbols turned up to absurdly loud levels.
The set of house keys jingled in front of a microphone does produce lots
of HF,
but we do not ever require the replayed recording of keys jingled to be
louder than if we slip a set of keys
out of a pocket and jingle them loudly as possible in front of us.

So the Quad-II does manage to make sufficient power to give what was
considered to be superlative sound in 1960
when used with ESL57.

One lone 300B, with 1/3 of the power of the Quad-II would not give
enough drive to ESL57
unless you like sitting very close to the speakers, then it becomes a
headphone experience.
So you'd have to use a quad of 300B in parallel SE to keep up with a the
Quad-II,
because the Quad-II does have a slight ability to make more
instantaneous AB power than the class A power
when tested with a sine wave, and at lower loads than the 16 ohms.

Quad-II has its Rout = 1 ohm, and hence the output voltage sags badly at
HF, about 3dB even at low levels when
driving 2 ohms at 15 kHz.
Peter Walker, who knew a shirtload more than I do, and a trouserload
more than Trever does
took this into consideration, and you will find the ESL57 will give a
near flat response driven with a 1 ohm source Z.
Many but not all SE amps with ZERO GLOBAL NFB are rarely ever going to
measure with Rout 1ohm,
so you get considerable sag in treble voltages with poorly made SET.
Some ppl quite like this, because if Rout = 3 ohms, the roll off at
10kHz with ESL57 still wouldn't
be too bad; many older people don't mind if all information above 10kHz
went AWOL.
But I still prefer a flat response.

But it definately is possible to make SE amps which achieve Rout = 1ohm
or less easily, something
TW would not know how to achieve, even if his life depended upon it,
because if he did,
he'd recommend how to get sufficiently low Rout correctly.

GNFB is one way to lower Rout, and you also lower THD/IMD,
and if the circuit is linear enough to begin with the added low level
harmonics created by
applied GNFB won't be audible.
Another way is to raise the anode loads with a higher than normal OPT
turns ratio.
But this reduces the power mabybe by 1/2 to 4 watts per 300B, and
although it sounds blameless, its expensive....

Unfortunately, we live in a marketing world driven mainly by greed and
BS.
So SE amp makers try to cut costs and boost profits by lying about the
capabilities of their Crud Production models being sold online to
gullible ppl.
Often the Rout is 3 ohms, way too high for any speakers.

A 300B is often loaded with a load for maximum power and symetrical
clipping,
and this load = (Ea / Ia) - (2 x Ra).
So for a given Pda of 32 watts for the 300B, one can select an Ea, find
out the Ia, and then the Ra,
and work out your load. Most makers choose 400V and 80mA and Ra =
800ohms,
so the anode load is 3,400 ohms.

An OPT is then made with a 3.4k : 8 ohm match.

This gives a 425 : 1 Z ratio.

And it means that where the triode Ra = 800 ohms, it appears as 800 /
425 at the OPT sec, ie 1.88 ohms.
Added to this is the OPT winding resistance, often 10% of the secondary
load value,
so total Rout = 2.88 ohms with no global NFB.
Adding just 10dB of GNFB will transform the Rout down to about 0.8 ohms
and the SET 300B will
be OK to drve ESL57, except for the volume ceiling.

So use another few 300Bs so you have 4 all in parallel, and reduce the
OPT Z ratio so each tube sees 3.4k,
then we'd have an OPT with 850ohms:8, or Z ratio 106:1, and the Ra of
all 4 tubes becomes 200 / 106 at the sec, ie, 1.88 ohms,
and with the same 10dB GNFB the Rout becomes 0.8 ohms approx or lower if
we use an OPT with
5% or lower losses, something few commercial makers ever try to do
because they hate
spending money on production costs, and the freight costs due to weight
mount up.

The amp with a quad of 300B will make a max PO = 32 watts, and it will
surely drive ESL57.

If the sec windings on the OPT can be re-arranged to offer a match to 4
ohms, the power will reduce
when 8 ohms is connected to about 20 watts, about equal to Quad-II,
but Rout will be 0.4 ohms.
Distortions of any kind at the 2 watt level won't be meaningfully worse
than the Quad-II.
We must remember that with two MATCHED KT66, and MATCHED EF86, and with
accurate R values in the Quad-II,
the distortion is well below 0.1% at 2 watts/8 ohms, although it is
mainly 3H.
But seldome are Quad-II amps seen with brand new matched tubes fitted
and I have often
seen such amps with very serviceable tubes one wouldn't bother replacing
for a few more years
that have 2H THD some 5 times above where it is supposed to be, ie,
below the
predominately 3H THD of all PP amps.

So if you had a 32 watt SE amp made with class A triodes, the THD
difference to
the real world operation of PP amps becomes purely academic, and of no
concern.
The bit of 2H and 3H in either types of amps doesn't ruin music because
its produced while the tubes concerned are all remaining well within
their class A linear region of operation,
and nothing is switching on and off.

I've just built the first of a pair of SE amps using 2 x 845, equivalent
to using 7 x 300B in parallel.
The sound is truly wonderful. It measures well.

THD in SET amps can be dramatically minimised by careful arrangement of
the driver tube
so that considerable natural 2H cancelation occurs; instead of 2H
CURRENT cancelation you get
in a PP amp, you can have 2H VOLTAGE cancelations in cascaded triode
stages.
The small amount of GNFB does the rest.

If TW wants to place ALL SE amps into the same category as being awfully
problematical,
let the world judge him as being plain wrong.

If one does own the most horridly awkward to drive Martin Logan ever
made,
then one has to think very carefully about what sort of
amp one uses. Ditto AR9 dynamics, which have low Z at LF, where you'd
want the Z to be highish.

One MUST do one's calculations about what maximum power is needed, what
max current,
and simply work out what class A current is needed through the output SE
devices
so that the amp doesn't clip when trying to make twice the max speaker
current.
My guess is that the Worst To Drive MartLogan probably needs 3 x 845 in
parallel,
which gives 75 watts of SET PO.
Frankly, 4 x 6550 well set up for a lot of class A and about 90 watts
max in PP
would also work and sound well, see my 8585 amp at my website.
I've also considered 2 x 845 in PP with the first 30 watts in class A;
should be awesome!
The 845 has Ra = 2k2, and a class A load is about 12k ideally,
so without GNFB there is a better ration of Ra to load than with a 300B.

Just rabbiting on about saying SET is a problem without
mentioning the conditions of use is just rabbit chatter, and it dirties
up the Internet.

Meanwhile, I had to repair an ME850 for the second time in two months
because
a thermistor went west on one channel giving a false indiaction of the
amp becoming too hot.

The ME is a powerful Oz made SS 200W amp with huge current ability, and
it does put up with horrid loads.
To my ears after repeated tests with it now working well, it sounds
better than
all generic budget Marantz, Nads, Denons, Yamahas etc etc.
But certainly no better than the 845 amps. Or better than an 8585,
or the ARC VT100 I recently re-engineered to my own circuit topology.

I suspect the amp played up because of a lightning strike. Lightning
does funny things to conplex
SS circuits for sure. Vacuum tubes giggle in their sockets during
lightning strikes.
Atomic bombs going off nearby don't affect them either.

But when the owner comes around to collect his repaired amp today
he tells me he pulled the covers off the bass drivers on his AR9s,
( bloody ancient old speakers!! ) he finds all the speaker surrounds are
broken and barely
holding the cones aligned.
So I could assume the voice coils will jam on their magnet poles with
any high level,
and when they do that they become a low Z, and this can well wreck a not
so good amplifier.
I told him not to use his speakers until he gets the surrounds fixed.

An SET amp with a lone 300B would certainly get lemon of the week award
if used
with an AR9 in good condition let alone in the poor condition I've
mentioned.

I then had a great laugh at the response curves printed out on the AR9
owner manual sheets the guy brought me for a look.
Two years ago I re-engineered a very damaged pair of AR9 and tested one
when i
got it back to original working condition.
The manual sheets show a very nice almost ruler flat response with
slight bass boost,
but what I measured with the AR9 I fixed was as flat as the US Rocky
Mountains,
with peaks and troughs up to +/- 8 dB along the band.
This is the typical result when I measure lots of nice brand speakers.

Anyway, looks like another re-engineering job for me in late autumn.
During the exercize I may include replacing ALL drivers, and completely
re-enginering a new Xover; the original AR9 is a horror story designed
by US bean counters.
Instead of appallingly low impedance at bass F of the original AR9
the reformed speakers will have a civilised and near constant Z and be
more able to be driven
by ANY amp, and one without having to have absurdly high voltage OR
current abilities,
And even an SET amp will cope quite well, if built properly.

Trouble is that whatever i do is limited by funds available....

Last time I did up a pair of AR9, I replaced the surrounds on the 4
woofers,
and used Peerless replacement minor drivers.
The final response was MUCH flatter, and the owner was rapt.
Drivers for speakers have improved vastly since the days when AR9 were
regarded well,
and I could say almost anything made by the North Europeans would be
better than anything made in the US in 1980.

Now George, down boy, no need to send those B52s out to stop me saying
what i think about
ancient old manufacturers in foreign powerful countries.

Patrick Turner.











Trevor Wilson