Well folks, it has been a while since I have written about Cables. So, if you don't believe in Cables, or if you have bought the latest boutique Wire and want to remain happy with your Purchase, you best stop reading now....
This "naked Truth about Speaker-Cables" could be seen
as second installment to "The Naked Truth about Interconnect-Cables".
It deals with a few more of the issues around Cables in general
and focuses on Speaker-Cables.
The DIY-part will attempt to provide a simple, inexpensive and
easy to make Speaker-Cable. It will nevertheless, due to its inherent
construction, provide a significant improvement over most commercial
Speaker-Cables and if I dare say so myself, even over the TNT-Star
DIY Speaker-Cable.
Anyone intending to save the dry theory and just wanting to get
stuck in with making cables; you can go ahead and make one out
of these two:
A Cable intended to be an all out Assault on the state of the
Art, the "UBYTE-2" Cable .
Alternatively you can keep it a bit simpler and more sane. Just
make the "FFRC" (Full Frequency
Range Cable). This is a Cable intended to be a significant
Upgrade over something like the common, moderately expensive,
multi-stranded Cable generally employed and sold as
"Specialist Speaker-Cable".
In more than a few cases I have seen Cables, were the "customisation" actually meant a Jacket in the latest Designer-colour and some fancy print. Some Companies do exist, which do a lot of fundamental research and often make some or all components for their Cable in their own Factories.
The numbers are however VERY few, as such an approach requires a lot of capital to be invested. Other Companies have found by accident or research a certain commercially made cable or wire which worked very well. They strated to market a cable based on this and later managed to have this Wire improved by adjusting it to their Spec.
In many cases I have found that the price cannot be taken as indicator
as to how and by whom the Cable you are buying has actually been
manufactured. So I think it is fair and true (but very uncomfortable)
to say that most specialist "HiFi-Cables" are an exercise
in marketing to about 95 % and maybein Research for 5%.
Please remember, not all companies selling specialist HiFi-Cables are
guilty of these practices, but many are. Please also note that in my
following Article I do mention a number of companies making or selling
HiFi-cables.
Their inclusion does not constitue as such an endorsement of their
products or indeed a statement to the contrary. It is simply that
their products are well known and often are typical for a specific
design-technique mentioned.
It should be noted that depending on external and interface conditions
second order effects and first order effects can sometimes change
places in terms of magnitude....
I would propose, that the usual RLC parameters (as with Resistance
[R] , Inductance [L] and Capacitance [C] ) should be viewed as
first order effects, though not in all situations each parameter
carries the same weighting.
Let's have an example for the different weighting in RLC Parameters, due to the relevant source and load impedance's. In Interconnects the source-impedance is between a few ohms and a few kilo-ohm and load impedance between 10 kilo-ohm and about 1 mega-ohm (both usually mostly resistive).
As a result the Capacitance (C) is a prevalent characteristic
with Resistance (R) and Inductance (L) relegated usually BELOW
most second Order Effects in their magnitude of sonic impact.
A second order effect, the Dielectric Absorbtion (DA) becomes
here a first order effect in audible magnitude. So the main parameters
for Interconnects are C and DA as long as R and L are kept in
sane regions (see "The naked Truth about Interconnects").
The Skin effect remains relevant (but firmly in the second Order
Camp) as does the Maxwell Effect (more on both later).
The limitation in bandwidth is mostly effected by the Lowpass
Filter composed out of the source-impedance and the cable's capacitance.
A Bandwidth of about 100 kHz is desirable for this interface to
make sure that the phase-shift and frequency response drop at
20 kHz remain acceptable.
The DA will determine time-smear and distortion of the Cable together
with further second order effect and the third order effects.....
Thus, unlike as in Interconnects, the Capacitance and DA can (mostly) be relegated into the second Order Camp. Due to the low impedance's in the load and the Source impedance of the Amplifier, the R and L of the Speaker Cable become highly relevant.
The Skin and Maxwell-Effects are being promoted from second order
effect to first order Status. The bandwidth of the Cable and the
frequency-dependent phaseshift will be a direct function of these
parameters.
The Situation is complicated by the fact that certain amplifiers
are very sensitive to capacitive components in the Speaker-load
(NVA, NAIM, Linn to name a few culprits). So with Speaker-Cables
indeed many bet's are off.
It seems that for Speaker-cables a low but matched R and L (so
that the attenuation remains constant with Frequency over the
Audio-band) is desirable combined with a low Capacitance and high
quality dielectric.
It is essential that the Skin-effect and the Maxwell-effect are
taken into account.
>From my simulations and practical tests it also seems desirable
to allow for an optional Speakerside "Terminator" Network.
This should compensate the Inductive Rise of the Tweeter if required
(depends on the Cables RL Values and the Speakers X-Over Design).
It may also be advisable to build into the Cable an (optional)
Pi-Network at the Amplifier Side. This will ensure the stable
operation of even the most "Hairshirt" designed Amplifiers
and prevent RF Ingress into the Amplifiers feedback loop (if any
is used) via the output.
The possibly most objectionable second order effect is due to the choice
of the conductor. Here we could be using multiple non-individually insulated
Conductors or the use of so called "solid core" cable.
A specific form of the Solid Core Cable is the so called "Litzendraht"
a braid made from individually insulated (enameled) conductors.
True "Litzendraht" (Litz-wire) is braided similar to
Kimber Cable and was originally invented by Nicola Tesla WAY BACK
in time. Modern realisations are often called "Hyper-Litz"
arrangements, why the "Hyper-Litz" I do not know.
Because these cables eschew the braiding and instead use simple parallel
wires, they loose the Litzwire's advantage of canceling the wires magnetic
field to a good degree....
Now let's look at the typical multistranded speaker-cable, which may be
the well-known "Monster Speaker-Cable" or the many clones sold
by anyone from Radioshack/Tandy to Wall-mart.
This uses a large number of non individual insulated copper conductors
twisted together for each Conductor. It is usually sheathed in transparent
or clear PVC or PU. The Geometry is the so called figure-8 pattern, also
called shotgun configuration.
Multistranded conductors have a problem. In an Ideal world, no electrons
would ever "cross" the boundaries between the individual
conductors. In the real world they do that all time.
(see Article Quantum Tunnel of Love, issue 8a,9a/92 of Bound for Sound)
Both the huge metal-to-metal surfaces themselves, crystal grain boundaries
and surface oxidation make the interstrand conduction much less linear than
conduction through pure copper.
In effect we introduce something not entirely unlike (but also
not entirely like) the crossover distortion of a Solid State Class
B Amplifier.
A solid copper conductor or a Litz-type wire will still have some
non-linear conduction due to impurities and Grain-boundaries,
but these are much less in magnitude.
Our next stops are Skin- and Maxwell- Effect.
The skin effect says that with rising AC frequency the electron
flow is being pushed more and more to the outer surface of the
Conductor.
It does not matter if a Conductor is 12 Gauge solid Copper, or a
Conductor of a 12 Gauge multi-stranded Cable ("Monster-Cable"),
both will appear as 12-Gauge solid Copper round conductors.
So, the higher the frequency, the more of the signal conduction will happen
in the outer layers of the cabel.
The conventional cables multistranded conductor will show even more
problems, due the non-linear interstrand conduction near the surface,
the presence of surface oxidisation and the like. Can you spell
Treble Grit....?
For all it is worth, the Skindepth for a round Copper Conductor
at 20 kHz is about equivalent to the Diameter of a 20 AWG Conductor.
At this depth from the conductors surface the current density
is 63 %. Hence a 20 Gauge conductor should not experience skin-effect
related problems below 20 kHz. As I have already mentioned the
100 kHz Bandwidth Requirement we should really expand this to
apply to the Skin effect.
It should be noted that the remaining current-flow between
this depth and the surface is heavily skewed towards the surface
of the conductor, so in more ways then one we'd rather like as
thin a solid conductor as feasible.
It appears that 24 to 26 Gauge individual conductors make for
a good compromise between bandwidth and manufacturing requirements
(or the ability to adapt readily available commercial wire for
Speaker-cables).
The Maxwell Effect works at the other end of the Spectrum (bass)
and is a bit harder to explain. I will not even try. Read the
Paper Prof. Malcom Hawkesford submitted to AES (Audio Engineering
Society) if you feel like doing a bit of serious mathematical
self abuse.
The Upshot is that a thin conductor will also IMPROVE the LOW-END
performance. Hence the Conductor providing the widest bandwidth
(measured and subjective) all else being equal is the thinner one.
A thin conductor introduces a lot of resistance, giving us problems
with the Series Resistance in our Cable.
So we to use for example flat, thin and wide Foil Conductors to
get the resistance down to a sensible level for Speaker-Connections
as implemented for example by Goertz Cable, Sonolith and Magnan Cables.
Another option is to arrange a lot of thin individual Conductors
in a Litz or "Hyper-Litz" pattern (XLO, Audioquest,
Cardas, Kimber and Tara).
Either solution provides us usually with some problems regarding
the Cables Geometry and hence often the RLC Parameters are shifted
in ways that are undesirable.
It will ideally employ multiples of very thin round Conductors
with individual insulation or use thin Foils.
It will not use multistranded Conductors and it will minimise
both Skin- and Maxwell- Effects.
The UBYTE-2 Cable, developed by me, conforms to most conditions.
An exception is that about one third of the Conductor-CSA (cross-sectional
Area) for each "leg" is made up from 18-Gauge Solid
round copper. This will restrict the Bandwidth of Cable slightly,
but helps to utilise a relatively easily available commercial
coaxial wire and to achieve a reasonably low DCR.
The outer copper Foil Conductor carries the major load of the
current and conforms to the solid and thin model.
The specific geometry (as developed by Jon M. Risch) allows for
a fairly good set of RLC parameters.
For a 5m Length of Speakercable there is about 0.1 Ohm DCR combined
with about 1uH Inductance. The Capacitance is around 800 pF for
5m.
Into a resistive 6 ohm load (respective of most modern tweeters
+ Zobel) load that will allow a - 3dB Bandwidth in excess of
300 kHz.
The maximum frequency response deviation over the 20 Hz to 20kHz
frequency range for a Speaker falling to a 4 Ohm Minimum will be
about -0.2 dB. This will be at the 4 Ohm minimum, as compared to
an infinite load impedance.
So a DC to 60 kHz bandwidth with a +/-0.1 dB deviation from the
1 kHz point and minimal phase response deviation should be possible
into a compensated Loudspeaker. The compensation may be part of
the Cable if required.
That is not very good, but I think it a tolerable technical
performance. In many cases the Output impedance of the driving
Amplifier will produce larger errors. Many expensive commercial
cables do not remotely achieve this standard of performance.
Also the FFRC - "Full Frequency Range Cable" is still very good with respect to fulfilling the requirements for high quality Audio.
The use of multiple individually insulated conductors of 24-Gauge
thickness guarantees freedom from the effects of non-linear conduction
as found in multistranded cables. The tickness is such that any skin-effect
related problems are pushed out of the audiorange.
With a high quality insulation and a geometry to minimise inductance
and capacitance even the lowly FFRC still is miles ahead of all
multistranded cables, regardless of make or price.
I have carried out an extensive series of Measurements and PSpice
Simulation that included popular Cables (like Kimber 4TC, Goertz
MI-2, Cable Talk 3 and Reson LS-350 and others). These confirmed
that both into a matched and an unmatched simulated Speaker-load
the "UBYTE-2" Cable will give the overall flattest
response.
The FFRC is not that far behind, but is solidly beaten by the
Goertz MI-2 and ever so slightly by the Kimber 4TC.
The rest of the Cables was just terrible.
A partial exception was interestingly the Reson LS-350, a very
thin Cable with a pair of widely spaced thin and solid conductors.
This cable has a high series-resistance and inductance. Most people
and reviewers take an instant dislike to this cable because of the
thinness of it's conductors. Yet it's bandwidth into a real-world
Speaker was surprisingly large.
However with its rather thin cross-section it will likely not
make a good match with quite a number of speakers, so as a universally
applicable speaker-cable, it is not too well suited.
Hence I think it can be said that the basic engineering for both
the "UBYTE-2" and the "FFRC" is sound and
fit for the purpose.
The fact that it outperforms on measurements and listening tests
almost any sensibly priced Speakercable on the Market earns the
"UBYTE" (Usually Beats Your Terrible Engineering) tag
for the "UBYTE-2" Speaker-cable.
In combination with (optional) Speaker- and Amplifier-Side Networks
we can match this Cable to almost any conceivable combination
of Equipment.
Various "audiophile" construction details (Air/Polyethylene
Dielectric, Solid Conductors, Foil Conductors and the like) address
many suspected but scientifically largely unconfirmed effects,
detrimental to sound quality.
The UBYTE-2 has yet to bettered by a commercially available Cable
in the Mid-Price Range up to at least 30-50 Dollar Meter. I
have not yet been able to try even more expensive cables in a
head to head test....
In comparison, the "FFRC" is only "good". Its main advantage is material cost below many of even the most basic "Speaker-Cables" with a performance more akin to serious "High-End" Cable.... It also is much easier to make.
The next time around I will be talking about the biggest gripe in High-End Cables.
Something few people of any remotely engineering related background
would put any Stock in.
I will be talking about Mains-Cables.... And Mains conditioning.
Until then, good tunes to all of you....
© Copyright 1998 Thorsten Loesch and TNT-Audio
How to print this article