In article , mick
wrote:
On Fri, 07 Aug 2009 17:31:43 +0100, Eeyore wrote:




I'm unsure about drawing any conclusions from graphs that start at 5x
the accepted maximum audible frequency. I hope Jim has included tests
on VHF coax as speaker leads too - it makes as much sense to me... ;-)

The problem is that some amplifier designs can be upset by having a load at
RF which does not suit them. The classical symptom is either sustained
oscillations in the region around a MHz or above, or bursts of oscillations
with particular audio waveforms. This can alter the audio behaviour. The
effects are measurable as well as audible.

Alas, the amplifier designer has no control over what loads the user
connects. And this will change with the choice and length of the
loudspeaker cables.

*If* your amp has much output above 100kHz then it is faulty and needs
looking at - seriously.

I agree. However some commercial amps *have* produced oscillations like
this with some loadings. For all I know, some still do.

And one of the points of the RF measurements is that it allows you to
determine the cable properties which you can then apply at audio
frequencies to assess what changes may occur *in* the audio band even when
the amplifier is stable and happy. So the measurements are useful - if you
understand why they were made and how to use the results. :-)

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

In article , Rob
wrote:
Jim Lesurf wrote:
Hi,

I've just put up a new webpage that provides some measurements on the
properties of a variety of loudspeaker cables. The page is at

http://www.audiomisc.co.uk/HFN/Cables3/TakeTheLead.html

It is an expanded version of the article published in 'Hi Fi News' a
few months ago.

Slainte,

Jim


FWIW it means very little to me. You seem to assume a correlation
between frequency, resistance and sound.

Not sure what you mean, I'm afraid.


Perhaps a paragraph or two on what you might expect any measured result
in the context of your measurments to mean?

The measurements and results serve two purposes.

1) The size of the peaks and dips in impedance will vary with the choice of
cable and end-load (speaker). Using 'open' and 'short' means loads with
impedances as high and low as you can get compared with the cable
impedance. So you can expect the results to give you a guide to which
cables give the highest or lowest peaks/dips for real-world loads. Hence
the results give a sign of which cables would be more risky with amplifiers
that are not unconditionally stable, or whose behaviour can be upset by RF
resonances, etc. In particular, sharp dips down to very low impedance can
be bad news for a poor amplifier. Hence useful as a warning.

2) You can use the measured impedances as a function of frequency to
determine the electrical properties of the cables. Choice of 'open' and
'short' here makes calculating the cable properties simpler, although in
principle any two choices of loading with significantly different values
would do. In the absence of a the amp having an RF problem these values are
most useful for telling you the cable series resistance and inductance as a
function of frequency in the audio band. (Yes, both values can vary with
frequency, although probably not by much in the audio band.) Combined with
the loudspeaker impedance, these series values change the frequency
response in the audible range. So the values determined from the RF data
tell you something about what changes to expect in the *audio* frequency
response. In particular, you want low series inductance and resistance to
minimise alterations in frequency response in most cases. The snag is that
*very* low inductance, in our universe, means *high* shunt capacitance
which can change the response from amps that have an output series
inductor. (Which I would recommend they *do* have.)

I've certainly known about all the above for decades. It was taken as
standard knowledge by people I've worked with. Although I guess some
audio-only designers may not know how the cable properties can be measured
using a VNA in this manner, but it isn't unusual in RF/microwave
engineering.

There are a couple of follow-on articles, that do look at this further, and
include simple techniques - like the use of a series inductor and 'zobel'
on the amp to help protect it against (1). That is a method I've always
used as it works neatly. But there are commercial amplifier designs that
*don't* do this, so are exposed to RF loading by the cable and speaker. And
the use of a series inductor may mean you'd have to be wary of ultra-low
inductance cables for the perverse reason that they have ultra-high shunt
capacitance.

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

In article 4a7d2929.352510515@localhost, Don Pearce
wrote:


The reason why Naim amplifiers don't is that they failed to consider
exactly this problem in their design. If they had just used Figure 1
(which you deride) they might have made an acceptable product.

I can't comment on any current designs by Naim as I've not measured or
studied them. I should also say that it is quite possible to design an amp
which is unconditonally stable without it having an explicit output series
inductor. However, that said, I did many years ago do bench measurements on
a Naim power amp. And, yes, it gave bursts of oscillations on audio
waveforms when used with a cable that had low series inductance.

So this problem is not simply a theoretical one.

In my years of design I discovered something important. It doesn't
matter what is the maximum frequency you intend to put through
something. Your design must encompass the maximum frequency at which the
active devices can produce gain (something like Ft). It is all too easy
to end up with an audio amplifier which is so marginally stable at 30MHz
that it can oscillate into some loads. When that happens, yes, there
will be sonic consequences.

Yes. This is my experience as well. Perhaps enhanced for having designed
systems for up to over 300GHz as well as for the relatively low audio band.
Quite interesting to find harmonics or out-of-band oscillations for these.
:-)

I also recall using 'RS' UHF modules that all oscillated at about 1.5GHz. I
guess the makers only used scopes and analysers that went up to about
1GHz...

It is all too easy to make an amplifier that looks OK on a test bench
connected directly to a test load - then find it bursts into oscillation,
or its other properties alter - when given some other load. I've also seen
this happen when someone was using an oscilloscope that didn't reach the
oscillation frequency. So the audio waveform became distorted, but with no
visible sign of the RF bursts until they tried a faster scope.

Must admit I am surprised that Eeyore seems to have missed this point.
Although I can appreciate that audio engineers may not know the techniques
used by RF and microwave engineers to measure something like cable
properties.

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

On 2009-08-07, Eeyore wrote:

John Phillips wrote:

On 2009-08-07, Don Pearce wrote:
On Fri, 07 Aug 2009 09:10:40 +0100, Jim Lesurf
wrote:

I've just put up a new webpage that provides some measurements on the
properties of a variety of loudspeaker cables. The page is at

http://www.audiomisc.co.uk/HFN/Cables3/TakeTheLead.html

It is an expanded version of the article published in 'Hi Fi News' a few
months ago.

No conclusions section there, but maybe as follows?

1. If you open circuit the cable at the loudspeaker end, it is better
if the cable is somewhat lossy, as this will prevent the quarter wave
Mod Z dropping to too low (potentially damaging) a value.

Alternatively, perhaps, that a well-designed amplifier will have about
2 uH of good quality inductance in series with its output to avoid such
a case becoming damaging?

Funny, that's very similar to the value I use. And it'll have a series R-C to
ground to stabilise the load the amp 'sees'. This technique has been known for
many decades. It is even used in long line-level drivers.

Yes - I agree. It seems we are in harmony about the need for an amplifer
to see a well-defined load at frequencies well above the audio band. Hence
my use of "good quality inductance" which needs to avoid self-resonance
at too low a frequency to maintain isolation for whatever the user throws
at the amplifier in terms of cable and loudspeaker.

I learned a lot from designing and building my first power amplifier.
I saw undesirable behaviour into the many MHz region whenever I failed
to pay enough attention.

So, I'm surprised at your reaction elsewhere. Even in the audio band,
loudspeakers can present impedances from near zero to high enough to
be considered infinite. Out of the audio band this gets no better,
from what I have seen.

So it seems to me that investigating loudspeaker cables with loads
from zero to infinity, and at frequencies well above the audio band,
is perfectly reasonable.

--
John Phillips

In article , John Phillips
wrote:


I learned a lot from designing and building my first power amplifier. I
saw undesirable behaviour into the many MHz region whenever I failed to
pay enough attention.

Indeed. In fact there are two stages to this.

1) The designer has to be able to establish if his bench design is
unconditionally stable or not. And if not, modify or change, to obtain
unconditional stability, without fouling the performance in some other way.

2) To then ensure that this will be true for commercial versions made with
components with a tolerance spread of values, slight alterations in
wirings, etc.

The worry here is the 'WW' effect. That of designs where a prototype
(published in Wireless World for example) worked fine for the designer. But
when many readers make 'clones' some of them oscillate or misbehave in use
due to changes in precise component values, wiring, etc. Hence the old term
'a WW design = a Worked Wunce design' to refer to this possibility. :-)

So, I'm surprised at your reaction elsewhere. Even in the audio band,
loudspeakers can present impedances from near zero to high enough to be
considered infinite. Out of the audio band this gets no better, from
what I have seen.

So it seems to me that investigating loudspeaker cables with loads from
zero to infinity, and at frequencies well above the audio band, is
perfectly reasonable.

There are two aspects of this that have concerned me. One is that I am far
from certain if all current/recent commercial designs are unconditionally
stable - particularly as I don't see signs that any reviews routinely check
this.

The other is the lack of any info on what speakers do above the audio band.
When you then throw in a variety of types and lengths of cables, almost
anything could happen in some cases.

I chose to measure an LS3/5A as I had a pair to hand. No idea what other
speakers do above the audio range. There seems to be zero data. I doubt the
makers usually know or care.

BTW Given Eeyore's reaction I'd suggest people read the previous two
'cables' articles in the series as that did cover some points. e.g. the use
of output networks. Although more about this and other factors will be in
later articles.

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Jim Lesurf wrote:
In article , Rob
wrote:
Jim Lesurf wrote:
Hi,

I've just put up a new webpage that provides some measurements on the
properties of a variety of loudspeaker cables. The page is at

http://www.audiomisc.co.uk/HFN/Cables3/TakeTheLead.html

It is an expanded version of the article published in 'Hi Fi News' a
few months ago.

Slainte,

Jim


FWIW it means very little to me. You seem to assume a correlation
between frequency, resistance and sound.

Not sure what you mean, I'm afraid.


There's a relationship between three things: frequency and resistance
(the things you plot) and sound. If one of the things change, another
one will change ('correlation'). On further reading you suggest a
relationship between frequency, resistance and risk.


Perhaps a paragraph or two on what you might expect any measured result
in the context of your measurments to mean?

The measurements and results serve two purposes.

1) The size of the peaks and dips in impedance will vary with the choice of
cable and end-load (speaker). Using 'open' and 'short' means loads with
impedances as high and low as you can get compared with the cable
impedance. So you can expect the results to give you a guide to which
cables give the highest or lowest peaks/dips for real-world loads. Hence
the results give a sign of which cables would be more risky with amplifiers
that are not unconditionally stable, or whose behaviour can be upset by RF
resonances, etc. In particular, sharp dips down to very low impedance can
be bad news for a poor amplifier. Hence useful as a warning.


Ah, OK, good. But is it *really* risky for any amplifier that doesn't
carry a cable recommendation tag? By risk I assume possibility of
component failure.

2) You can use the measured impedances as a function of frequency to
determine the electrical properties of the cables. Choice of 'open' and
'short' here makes calculating the cable properties simpler, although in
principle any two choices of loading with significantly different values
would do. In the absence of a the amp having an RF problem these values are
most useful for telling you the cable series resistance and inductance as a
function of frequency in the audio band. (Yes, both values can vary with
frequency, although probably not by much in the audio band.) Combined with
the loudspeaker impedance, these series values change the frequency
response in the audible range. So the values determined from the RF data
tell you something about what changes to expect in the *audio* frequency
response. In particular, you want low series inductance and resistance to
minimise alterations in frequency response in most cases. The snag is that
*very* low inductance, in our universe, means *high* shunt capacitance
which can change the response from amps that have an output series
inductor. (Which I would recommend they *do* have.)


It'd be nice, although I expect quite difficult, if you could explain
how these effects could influence sound.

I've certainly known about all the above for decades. It was taken as
standard knowledge by people I've worked with. Although I guess some
audio-only designers may not know how the cable properties can be measured
using a VNA in this manner, but it isn't unusual in RF/microwave
engineering.

There are a couple of follow-on articles, that do look at this further, and
include simple techniques - like the use of a series inductor and 'zobel'
on the amp to help protect it against (1). That is a method I've always
used as it works neatly. But there are commercial amplifier designs that
*don't* do this, so are exposed to RF loading by the cable and speaker.

Sounds daft. Do you know which amplifiers?

And
the use of a series inductor may mean you'd have to be wary of ultra-low
inductance cables for the perverse reason that they have ultra-high shunt
capacitance.


Excellent! It's clear that the technically literate here know what
you're driving at, and if Maplin's own is good enough for you etc :-)

Rob

"John Phillips"

So, I'm surprised at your reaction elsewhere. Even in the audio band,
loudspeakers can present impedances from near zero to high enough to
be considered infinite.


** Fraid that is absolute crap.

Only a FAULTY speaker exhibit shorts or opens in the audio band.


So it seems to me that investigating loudspeaker cables with loads
from zero to infinity, and at frequencies well above the audio band,
is perfectly reasonable.


** Only if you are a pseudo academic, audiophool lunatic.

Cap fits you OK.



..... Phil

In article , Rob
wrote:
Jim Lesurf wrote:

FWIW it means very little to me. You seem to assume a correlation
between frequency, resistance and sound.

Not sure what you mean, I'm afraid.


There's a relationship between three things: frequency and resistance
(the things you plot) and sound.

All the plots show relationships between frequency and apparent load
'resistance' (actually magnitude of impedance). But that isn't directly
related to 'sound' as such.


If one of the things change, another one will change ('correlation'). On
further reading you suggest a relationship between frequency, resistance
and risk.

As above. Sharp/deep dips in the 'resistance' as you change frequency tend
to give more 'risk' that the amplifier will be affected in a significant
manner. But this isn't a simple relationship with 'risk' as that depends on
'risk of what' and choice of amp, etc. Decent amplifier designs will be
essentially unaffected by all this. But some amps might be unhappy.


Perhaps a paragraph or two on what you might expect any measured
result in the context of your measurments to mean?

The measurements and results serve two purposes.

1) The size of the peaks and dips in impedance will vary with the
choice of cable and end-load (speaker). Using 'open' and 'short' means
loads with impedances as high and low as you can get compared with the
cable impedance. So you can expect the results to give you a guide to
which cables give the highest or lowest peaks/dips for real-world
loads. Hence the results give a sign of which cables would be more
risky with amplifiers that are not unconditionally stable, or whose
behaviour can be upset by RF resonances, etc. In particular, sharp
dips down to very low impedance can be bad news for a poor amplifier.
Hence useful as a warning.


Ah, OK, good. But is it *really* risky for any amplifier that doesn't
carry a cable recommendation tag? By risk I assume possibility of
component failure.

It is certainly possible for an amplifier to exhibit uncontrolled RF
oscillations, and for those to then damage the amplifier. Possibly also the
speaker. But I can't tell you any value for the 'risk' of this happening as
it would depend on things we don't know. More likely is that the audio
behaviour may be affected without the amp failing.

Again, well designed amplifiers aren't at any 'risk'. If the designer knew
what he was doing they will be stable into any load.

[snip]

It'd be nice, although I expect quite difficult, if you could explain
how these effects could influence sound.

Again, depends on the circumstances. High cable series impedance will alter
the frequency response in ways that depend on your choice of speaker.

High cable shunt capacitance may affect response if the amp has a high
output impedance (very low 'damping factor').

But the details will depend on the specific case. The alterations may be
too small to be bothered with, or not...


There are a couple of follow-on articles, that do look at this
further, and include simple techniques - like the use of a series
inductor and 'zobel' on the amp to help protect it against (1). That
is a method I've always used as it works neatly. But there are
commercial amplifier designs that *don't* do this, so are exposed to
RF loading by the cable and speaker.

Sounds daft. Do you know which amplifiers?

I can't comment on any current or recent commercial designs as I've not
measured them, and reviews generally ignore this area. So no data. I think
it likely that most (indeed almost all) are fine as this should be a known
problem, and engineers determined how to fix it decades ago. Maybe they are
all fine. But... no data.

However I do tend to get an uneasy feeling when reviews ignore issues like
this for decades. It can mean eyes are not on the ball and problems
familiar to past generations of engineers may end up in new designs because
no-one is alert. I confess I do wonder when I see some of the more 'quirky'
designs sold at high prices that have all kinds of of characteristics.

I can only say that I've personally seen such effects in amps many years
ago. e.g. in the Naim amps of some decades ago. It is a common problem with
experimental designs which the designer then has to iron out.

The problem here is that it can make good sense to choose loudspeaker
cables with very low series resistance and inductance, but that this means
high capacitance with minimal damping losses, and unless the amplifier is
happy with this there may be drawbacks.

In our universe, the product of series inductance and shunt capacitance for
cables is limited by the speed of light. Lowering one tends to shove up the
other. To avoid this, invent warp drive, or use a wormhole in space for the
cable. ...or just keep down the length of cable needed. :-)

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

On Sat, 08 Aug 2009 10:17:24 +0100, Jim Lesurf
wrote:

It is all too easy to make an amplifier that looks OK on a test bench
connected directly to a test load - then find it bursts into oscillation,
or its other properties alter - when given some other load. I've also seen
this happen when someone was using an oscilloscope that didn't reach the
oscillation frequency. So the audio waveform became distorted, but with no
visible sign of the RF bursts until they tried a faster scope.

There is a general rule in design that everything will oscillate. The
only consistent exception to this rule occurs when designing an
oscillator.

d

Phil Allison wrote:
"John Phillips"
So, I'm surprised at your reaction elsewhere. Even in the audio band,
loudspeakers can present impedances from near zero to high enough to
be considered infinite.


** Fraid that is absolute crap.

Only a FAULTY speaker exhibit shorts or opens in the audio band.

How about this, the only impedance curve on Trevor's site?
http://www.rageaudio.com.au/index.php?p=1_12

--
Eiron.