John Stewart wrote:

Patrick Turner wrote:

John Stewart wrote:

Kevin Killebrew wrote:

Here is my question:

I've got a tubed receiver (Sherwood S8000) that is going to be
connected to
two sets of speakers via an external switch. One set is outside
on the
patio, the other inside at the bar. Most of the time only one
set of
speakers will be playing, but occasionally I forsee both sets
being played
simultaneously. Both sets of speakers are 8 ohms so when I
occasionally
run both pairs I will have four ohms (assuming they are
paralleled) per
channel. Which receiver output should I run to the switch
box? Four or
eight ohms?

Is it best to run an occasional four ohms on the eight ohm tap?
or run
mostly eight ohms on the four ohm tap having a four ohm load
only when both
sets of speakers are playing? I suspect the eight ohm tap is
the right
choice because it will be correct most of the time, but I wanted
some RAT
opinions of the effect of running mismatched impedances, i.e.,
four ohm load
on an eight ohm OT tap. I want to avoid an overly complex
switch
arrangement.

Kevin Killebrew
Austin, Texas

Whatever you do, don't wire the speakers in series.For a two
speaker hookup,

that removes the amplifier damping from both speakers since they
would be then
be sourced through each other, an intolerable situation.

The DF issue is not always important

We can see that as some folks still choose to use pentode output amps
with no voltage NFB. For example, it may work well for some musicians
in order to get the sound they want.

Many old radios used a 6V6 with no NFB.
Above a very low level they sounded attrocious.

Better than no radio.

And sure, with muso amps, DF doesn't matter much but many guitar amps
do have some NFB.



You can have 10 speakers in series and if each has the same
enclosure,
and each driver is the same, the response will be identical to a
single
driver driven with the same low Rout, and this proves the
amp damping doesn't reduce.

Within the limits of the similarities of the enclosures & drivers,
true enough.

I didn't mean it any other way.

As well as the minor differences between enclosures & drivers, in
actual practice there will be a spread of response since the boxes
will all have to be in physically different space.

Nope, you can have a line away of speakers and they all share the same
box.
The dozen+ drivers used act like one big long driver, with a dozen times
the
box volume.
You could even replace all the cones with a single long
lightweight diaphram glues to the voice coils, and curved in one
direction,
like a pipe cut down the centre, and with a suspension rubber along the
edges and at the ends.
That would make all the drivers improve their team work.


As a group they do not constitute a point source.

That doesn't matter to the builders of large line arrays.

So each speaker it it's box will have a unique response curve, unlike
any of the others. Your statement doesn't prove anything. It simply
makes an assumption & a poor one at that.

You are not proving anything either, except you are giving a lesson in
side tracking.
Muliple 2 way bookshelf speakers can be hooked up in series without
changing the near field response of each or how they operate.
They just have to be the same brand, model number, and fairly well
matched.
Its routine to make speakers closely matched.
Such matching gives the best imaging where a stereo pair is used.
But we are talking about DF and whether or not is is decreased, ie, DF
number
is lower, just because we series connect speakers.
Clearly I have made my point that if one connects a dozen
of the *same* bookshelf speakers in series, the DF is actually improved
if the same amp with the same Ro is used.


In fact it increases.
If you have Ro = 0.5 ohms and a 5 ohm speaker, DF = 10.
If you have 10 series speakers of 5 ohms for 50 ohms, same Ro, DF =
50.

In your example the DF for the group is 50. However, for each speaker
the source impedance would be the amp Ro of 0.5 ohms plus the entire
series complex impedance's of the other nine speakers in the group.
That is a condition I would prefer to avoid.

So would I, so line array builders connect groups of say 4 speakers in
series,
so 4 give 32 ohms, then they might have 4 groups of 4 in parallel,
pulling the
Z to 8 ohms, and the same as one single driver.
DF then is the same as one driver.


An even more important consideration is, how will we deliver power to
the load? In your example the load reflected to the output tubes (or
transistors) is now 10X nominal. So power available is drastically
reduced. As well, if a pentode output, distortion goes way up due to a
gross mismatch.

See above.

When multiple drivers are seriesed and paralleled for the correct Z
wanted for a good amp
match there are no problems.

If somebody wants to ties multiple speakers together more tightly
rather than hooking them up in series and expecting them all
to behave with the same parameters then I suggest
one use a tapped auto transformer with say 4 equal windings in series
and one could have 4 x 8 ohm speakers across each winding for 2 ohms,
then with 4 windings loaded by 2 ohms each you have 8 ohms input
for 16 drivers, and they are all locked together magnetically.






People who build line array speakers using a dozen or more drives
know
what I am saying to be true, or else they would be compelled to
connect all speakers in parallel, giving input impedances of very
low value.

Not at all. Systems are easily built using series/parallel connexions.

Ah, but as I said you have to series speakers in line arrays.

A set of four 5 ohm speakers is still 5 ohms when series paralleled.
The same is true for 16 speakers & these kinds of things have been
built by their fans for many years. I recall well a friend building
something called the 'Sweet 16' more than 45 years ago. It used a set
of sixteen 6X9 speakers in a single enclosure. It sounded pretty good
too. In a recent magazine article I see a project using four speakers
per enclosure, again connected series/parallel in order to maintain a
useable impedance level.

So, what's wrong with seriesing speakers?





However, where you have two different brands of speakers, different
boxes
and drivers, series speakers will produce big variations in response
from each
because the impedances of each are very dissimilar at different F.

That is what I recommended against in the first place. Why do we need
to hear that again from you? I certainly don't need your confirmation.

So its for this reason you wouldn't want to series speakers.

Set the amps for 4 ohms where you want to have a pair of 8 ohms in
parallel.

Pentode amps are a bit queer.
They have a load optimum where thd is lowest and each side of that
thd
tends to rise and thd is very high when RL is very high.

True enough. But how does that make a pentode a 'bit queer'? It is
simply a property of pentodes, which can be exploited. That is why I
recommended using the 8 ohm tap. By paralleling the load resulting in
a mismatch on the low side, the 3rd harmonic is reduced. If we were to
go the other way as you have recommended, 3rd harmonic instead goes
way up. The choice IMO is obvious to anyone who carefully considers
how a pentode works.

But if you palce two 8 ohms speakers in parallel across a 8 ohm outlet
from a pentode amp,
RLa-a is halved, and all the measurements i HAVE EVER CONDUCTED indicate
thd rises heaps,
and at high sustained levels the tubes will glow red hot and expire.

The gain of pentodes is about proportional to their loads, so if load is

halved, so is the output tube gain and so is the amount od applied NFB
so the thd rises because of this fact and the fact that the lower load
increases thd anyway.
I am suggesting ppl try to keep the amp working with its
recommended correct RLa-a, ie, with a pair of parallel 8 ohm speakers
= 4 ohms, and connected to a 4 ohm outlet if there is one.

But if ppl connect 8 ohms to the 4 ohm outlet, the pentode AB amp
will make marginally less maximum power, and its output tube gain will
near double, and the slight increase in thd due to a higher than ideal
load match
will be offset by a near doubling of the applied NFB.

Applied NFB depends on open loop gain
Without any load, and with NFB, pentodes don't have an alarming
increase in thd.






But the amp NFB also becomes most effective when RL is high, when
the tube gain is high, and also the PP pentode amp works in class A
when RL
is high, when it isn't producing power in class AB where RL is low.

When the Rl is higher (or lower) than optimum in a pentode amp,
distortion increases. Than more FB is required to get the distortion
down to a reasonable level. In a PP amp the even order (2nd, 4th,
6th...Etc) harmonics are reduced by the circuit. All the other noise &
distortion products are still in there.

But as I said, the NFB is increased hugely with the worst case
situation, no load.

A properly designed pentode amp with 20 Db of NFB only has 20 dB
of FB when RL = the rated load, say 8 ohms.
output tube gain is perhaps 25. But with no load at all, tube gain leaps

to near µ for the tubes at mid frequencies, so gain = 120 for EL34.
So the applied NFB with no load leaps about 5 times, or 15 dB to 35 dB.
So the thd is controlled.


In all the ordinary amp topologies, voltage NFB is applied around the
entire amp to reduce the D. However, by doing that there is a danger
of manufacturing some intermodulation products that were not
originally present.

IMD is more likely to be made in amps with low NFB due to NFB
application.
Where NFB application is high, there is less IMD, period.


NFB of this kind does not fix the D in the output stage.

Yes it does, where the tubes are not driven into clipping, and have
enough gain and are not
driven into grid current.

It simply tries to correct what is already there while at the same
time it is causing other problems. Better no miss match in the first
place.

Sure, but speakers present loads which are always going to provide a
mismatch
at some F.



This is an idiotic statement, but speaker Z varies for different F,

What is idiotic about that? It is a set of complex impedances, a
property of the speaker in it's enclosure & dependent on the space it
is in, as well.

and either its working in AB or it isn't, but one has to
contemplate that the amplifier is
sending a voltage x current signal to a constantly changing load
value.

Since its almost impossible to think of 3 things at once, I leave
you to contemplate the tube's operation knowing that
to win a game of chess against a naked woman and keep an erection
and do your tax return all simultaneously is downright impossible.

What has any of that got to do with the original query? Or is it
simply another 'Fog of Bull****' to cover something else you are
unsure of?

Speakers present a dynamically varying load with the signal
whose frequency content varies dynamically along with amplitude.

Should one contemplate what is happening at their amp's output
and within the amp, one should remember that the load is not a
resistance of one value.
Because all speakers supply a degree of mismatch, their power ceiling
is somewhat restricted below the ideal resistance value that gives the
ideal power.

Series speakers are OK if well considered, but not if they
differ widely in specs.

Nearly all pentode power amps are set up to cope with twice the
average rated load value quite well, but certainly not with half the
recommended
load value.


Better a lower Z match than higher since PP pentode amps benefit
from that. The
3rd harmonic is reduced. Most speaker loads are far too high
anyway at their
resonance(s). A parallel hookup tends to smooth that out provided
the speakers
are dissimilar.

Using two brands of speakers in parallel with different box
resonances at LF
will indeed tend to produce a more even impedance.
But also its likely that at some F the impedance will be lower then
4 ohms.

Some makers have deliberately different enclosure volumes and Fb
for a pair of bass drivers in the one speaker cabinet.
It makes the resonant Z peaks overlap, and a much flatter
Z is realised.

And easy to simulate for anyone with simulation software.

Its all in RDH4, with a graph.

And many other references, as well.

If one thinks about what is happening here, this whole exercise (the
debate) is ridiculous. Kevin K asked a simple question & several
responded with straight forward & too the point answers. Kevin told us
the speakers are in different locations ( And probably different
speakers). As one of my points, I simply recommended against a series
connexion for the speakers.

Let us grant Kevin that he has adequate intelligence to draw from what
we have said
what he needs to know.

The conclusive simple recommendation is series speakers are OK
if they are exactly the same. This won't hurt the amp.

If the speakers are different, paralleling them is all you should do.
But the combined nominal load may be too low for the amp
and cause heat stress, unless you can move the speakers from the
8 ohm connection point to the 4 ohm connection point if there is one.

Other matters raised about series/parallel speakers are just
salient matters raised, and should not distract from the
simple answer.



But our Patrick T knows better & proceeds to go into another of his
many long winded explanations, much of which was completely unrelated
to the query. Worse still, parts of his response are in error. It
seems Patrick T is for some unknown reason is driven to respond to
every post that shows up on the NG. He is far from the expert he
pretends to be. Others should be aware of that. Be sure to get a
second opinion!

But I don't recommend paralleling speakers just to get the 3rd harmonic
down,
which won't occur as you claim, and I won't recommend loading an amp
adversly with a lower impedance load that it was designed for.

Feel free to get as many opinions as you want.

I am not worried by 100 different opinions.

And at least I take no notice of your consisten complaints about my long
posts.

I reserve the right to talk around and through a subject in detail,
and if you don't like to have to read what people post here, then
you are not interested in details, and the nitty gritty.

This is a discussion group, and nobody has the right to
tell me how much I should say.

If its awkward for you to post an answer then so be it.

And BTW, I don't reply to every post, that's another thing you got
wrong.

Patrick Turner.




John Stewart

Patrick Turner.



Good Luck, John Stewart