Good, Neutral, Reasonably Priced Cables?

sean

are the zobel component values (R&C) cited for the goertz cable a function of its length (10 to 20ft in the tests), or can one use those R&C values for any length of goertz cable?

thanks for the most informative posts. the audioholics read was exceptional.

rhyno

This is unquestionably THE best thread since the censorship wars. I don't know how to read, but I listen carefully...

Since I missed the censorship thread, I don't have a point of comparison, but this thread has definitely been very informative and entertaining regardless.

Anyways, more info for all on ribbon cables, subject du jour that they are-

Magnan website 'white paper' on ribbon cable design-

Magnan info

Basically, some of the points here exactly mirror Sean's comments above re. skin effects-

"The skin effect phenomenon has been found to be the major signal degrading effect in conventional audio cables. These effects include smearing of musical details, smearing together of instrumental images, flattening of the sound stage, and usually a general overbrightness. Almost all conventional audio cables utilize relatively thick stranded or solid wires which inherently cause gross audio band skin effect time smearing."

From there, the points seem to diverge, and the white paper becomes a mixed bag that includes fun 'marketing' metrics such as the rigorously defined 'Audio Figure of Merit' in Figure 1. :-(

All that aside, if you check out the soundstage review at

Soundstage Review

the comments on the coherency of the system sound are pretty similar to those I made upthread.

Also, there is a comment on cable theory references at the Silversmith Audio site (another manufacturer of ribbon cables)- if anyone knows more about these, I'd love to hear about it. If not, we'll have to wait for Jeffrey Smith to update the site.

Silversmith

I'm referring specifically to this quote in the 'cable theory' section-

"In the last couple of years, impressive scientific studies have been conducted which have measured some differences in wire performance, including directionality, lending some credence to the subjectivist's camp. While the debate rages on, it is interesting to note, that the engineering knowledge needed to explain exactly why cables do make a difference, and accurately predict what a particular cable design will "sound" like, has been available for decades. Unfortunately for audiophiles, it was not until as recently as 1985 that someone actually applied that knowledge to the world of audio cabling. To this day, the Essex Echo - Unification Tracks 1-4, by Malcolm Hawksford, remains the single greatest work on the subject of audio cabling."

Sean, once again, thanks for the detailed post. I'm busy trying to break down your comments into digestible chunks for my 'challenged' brain. My first question regards your comments on 'minimizing skin effects'.

I guess I don't understand all of the relevant length scales that come into play, so I'll think about this from first principles- if a 'skin effect' is always confined to an esentially infinitely thin skin (on the order of 10s of nanometers; i.e., a few hundred atoms of thickness), then I have a hard time understanding how cable geometry matters at all.

In this case, if one thinks about a cable with a circular cross section, then basically the circumference/area of the cross section is the 2-d analog of the surface/volume ratio. Cicumference/area is always 2/r (r = radius.)

The geometry of a ribbon cross section isn't much different- so long as a ribbon's width is relatively large in comparison to its thickness, its perimeter/area ratio approximates 2/t, (t is the ribbon thickness). This approximation holds pretty well for both Alpha-Core products (MI-2 is w = 0.75 inch and t = 0.01 inch) and Magnan products (Reference speaker ribbon is w = 1.25 inch and t = 0.00075 inch).

What this means is that if skin effects really are confined to a very thin layer, it is immaterial whether cable cross section is circular or ribbon-like. For any given radius r = thickness t, the surface/volume ratio is the same.

Taking it a step further, in comparing surface/volume ratios of any two conductors (a and b, lets say), the ratio of the surface/volume ratios (SVRa/SVRb) is rb/ra, or, if a is a ribbon and b is a wire, rb/ta. If 'minimizing skin effect' is equivalent to minimizing surface/volume ratio, then basically the thicker conductor wins in this scenario, regardless of cross-sectional geometry. This outcome seems counterintuitive given everything one sees in cable design.

Still with me? Yeah, me neither...

What I think must actually be going on is that the 'skin affected zone' is relatively deep (let's call the depth d) compared to r or t in any given conductor cross-section. In this case, one can break the conductor down into outer 'skin affected (sa)' and inner 'bulk (b)' regions. Deriving geomeric relationships between these regions must yield some difference in the behavior of the ratios of 'skin affected area' to 'bulk area' for the two geometries (circular vs. ribbon.)

I'm too lazy/tired to do the math at this point- if someone could confirm that I'm either going in the right direction, or completely lost in the woods, I'll be more motivated to revisit the problem later. On the other hand, if someone wants to pipe in and keep me from reinventing the wheel in this analysis, that would be great too.

Next up- thinking about phase errors...

Flex: A good yet "basic" power supply would consist of parallel RF bypasses across the incoming AC line, an EI type transformer ( NOT a toroidal ), fast recovery diodes ( or snubbers across standard diodes / rectifiers ) and a staggered array of multiple value filter caps. None of this is that hard to do. Just these things in itself would be a big step forward for most designs.

Quite honestly, toroids are crap compared to a well designed EI type transformer and that's why i specifically stated the old "iron core" type transformer. Since toroids are much cheaper than an EI, guess what most mass produced and even the majority of "high end" gear uses??? That's right. The cheap junk that is touted as being " a technological advancement".

In case you think that i'm making this up or are wondering why i said what i did about the toroidal type transformers, a really good toroidal will offer about -85 dB's of high frequency isolation from line noise. Some would consider this "excellent" and more than enough. That's because they are used to working with spec's that have been shoved down their throat as being "acceptable" by the industry and have come to believe them to be "as good as it gets". Not even close.

When you compare this to a really good EI type transformer, you're looking at an isolation factor of appr -145 dB's. In plain English, the difference figure between the two transformers is -50 dB's of attenuation. This means that the toroid could potentially allow 60,000+ times more noise through than what the good IE "iron core" type transformer would AND it would still be doing its' job "as expected". Now can you folks understand why i said that toroidals are crap???

While some may doubt the figures that i've quoted here, do some research. As far as the validity of the -145 dB figure on the iron core, it is achievable. If you doubt this, ask Larry aka LAK. He's using some transformers that have this spec that i helped him locate for his AC filtration system. You might also want to look at some of the comments that John Curl & Bob Crump of CTC Builders have made pertaining to toroids vs iron core's. That is, they have both flatly stated that toroids offer nowhere near the isolation / noise filtering capacity that an iron core does. This is besides the distinct advantage that they have in terms of low frequency "punch".

If one wanted to take that all of that a step further, you could install a zobel network to reduce the ringing of the transformer and lower the noise floor and / or install some type of low-pass filter. This would require a very specific orientation of the AC plug for proper operation since the energy that was "trapped" by the filtering would be shunted to ground. If one wanted to really get adventurous, they could build a resistive trap rather than shunting it to ground. Only problem is, you have to use resistors that can dissipate enough power and provide heat-sinking for them. Once again, this raises the cost of production and increases the complexity of the design. You'll NEVER see anything like a resistive trap in an audio circuit though. The primarily reason is that it costs too much and the second reason is that most audio engineers have never seen or heard of such a design. Maybe back in their textbooks or in school, but never in the real world.

What would make this even more effective would be to use the chassis as a "Faraday shield" i.e. where the chassis completely isolated from the circuit path and is tied to Earth ground. Many circuit designs tie the chassis into the circuit path, which is phenomenally stupid as far as i'm concerned. This is done because it is FAR cheaper and faster as far as production is concerned, so the bean counters tend to like these type of cost-cutting production short-cuts. Removing this from the design means a lot more point to point wiring and / or increased complexity of the circuit board design. Since one means more labor and the other means more parts, the bean counters don't like that approach.

As far as you question goes about cable resonances and damping, my specific comments were based on a relatively popular cable in use. I have to assume that since people are using this design, there are other similar designs on the market too. As i mentioned, this cable is VERY rigid even though it makes use of stranded conductors. If you were to conduct simple tests using both your hands, you would understand where i'm coming from. That is, you could "flick" the cable at one end and literally feel, let alone hear the "thunk" at the far end. NO actual measurements are needed as the results are blatantly obvious.

As far as clarifying what i meant by "jacket", i didn't mean the dielectric material surrounding the conductors. To me, you have the conductor ( copper, silver, etc... ) and then you have the "dielectric insulation material" that sheathes the individual conductors. All of these insulated conductors are then housed in one larger container, which i'll refer to as the "jacket". The jacket simply acts as a container for all the various insulated conductors.

While the jacket is also a dielectric, i was trying not to confuse the issue between the jacket and the individual insulation for each strand of conductors. As such, using a cable with a very soft i.e. "rubbery" jacket can very definitely reduce the amount of vibration that is allowed to travel from one end of the cable to the other. I have a near identical design to the "very rigid" cable using the same gauge conductors, but with very different dielectric around each conductor and with a different type of "jacket" around those. The differences in how much mechanical energy that can be transmitted through these cables from end to end in a side by side test is rather amazing.

On top of all of that, the "mechanically lossy" dielectric material mentioned above is typically very good at absorbing higher frequencies. This tends to reduce the bandwidth of the cable and act as a passive filter in itself. As such, the use of "low loss" dielectrics ( like Teflon ) in a power cord is backwards as far as i'm concerned. That's because Teflon is both more rigid AND it is of lower DA ( Dielectric Absorption ) than many of the other options available to us for a project like this. With AC, you do NOT want wide bandwidth, you want very narrow bandwidth. Since damping mechanical resonances AND increasing the DA ( Dielectric Absorption ) of the power cord i.e. limiting the bandwidth can be achieved using softer, more "rubbery" types of insulation, you get two birds with one stone. On top of that, these materials are both cheap and plentiful, so going any other route is both senseless ( as far as i'm concerned ) and economically wasteful. This is NOT true for signal cables though, so don't think that cheap dielectric is "good" for speaker cables or interconnects. After all, power cords are dealing with a 60 Hz signal whereas music is generally considered to be from 20 Hz to 20 KHz.

Obviously, there are going to be a LOT of cable manufacturers that charge outrageous sums for their fancy power cables upset with me and wanting to disagree with this observation. As such, i'm more than open for debate on this subject. Maybe we can even get "Audioholics" to perform some "third party" tests on various AC cables and see who's right on this one too : )

Rhyno: There are different ideas as to what values should be used. Changing the values will not only affect the "hinge frequency" that the Zobels start acting as part of the load, but it will also alter how effective they are at damping reflections. Much of this deals with what is called "transmission line theory", which so-called "cable experts" say does NOT apply to audio circuits. Personally, i think that transmission line theory DOES apply to audio in many ways and that may be why my thoughts / beliefs about various cables don't fall inline with most "experts" on the subject.

As you saw in the tests, when Audioholics changed the values used for the Zobel's, this also changed how linear / in-phase the signal was at both ends of the cable and how much of a reflection they were able to measure. To be fair though, ALL of these "problems" that the Zobel corrected were well into the MHz range, which is measurably beyond what most people are using for audio amps. Then again, with all of these concerns about RFI entering the system, why would you want to use a cable that was capable of introducing RF based ringing directly into the system when all you would have to do is to use a few parts to make a Zobel with? This is yet another reason why i've stressed the importance of a Zobel with very high capacitance / wide bandwidth / low impedance cable.

My theory about Zobel's is a little different than most others. In the above article i mentioned that Nelson Pass wrote about speaker cables, he mentions a specific set of values that he likes to use and recommends for use with "low inductance" speaker cables. In that same article, Nelson Pass quotes Matthew Polk as suggesting a different set of values for the same cables. If you ask Jon Risch what values to use, he'll give you figures that are somewhere between what Pass, Polk and Goertz use. To be honest, they are all effective formula's, but some may be more suitable for specific designs than for others.

How a Zobel works is that you have two parts i.e. a capacitor and a resistor. These are wired in parallel ( across ) the circuit i.e. at the speaker terminals across the positive and negative binding posts. What happens is that at a certain frequency, the capacitor will start to conduct signal to the resistor. Below that frequency, the Zobel is basically "invisible". Once we hit that frequency, the resistor acts as a "dummy load" or "signal absorber", presenting the amp with a purely resistive i.e. non-reactive load. By changing the value of the cap, you change the frequency of where the Zobel starts to work at and as you change the value of the resistor, you change what impedance the amp sees above that frequency. As such, it is VERY important to use "non-inductive" resistors as part of the Zobel, as inductive i.e. "wire wound" resistors may not be of wide enough bandwidth to work properly.

This is where it gets tricky and why there are different ideas about what values to use. Since some amps are more / less stable than others, some folks want the Zobel to come in very quickly i.e. at a lower frequency. This can definitely increase the stability of the circuit, so some would consider this a benefit. Other folks believe that you want to avoid ANY interaction between the audible range and / or any of the harmonics of the extreme treble range, so they want the Zobel to come in at a much higher frequency. As such, each individual selects a capacitor value that reflects the frequency that they want the Zobel to start conducting at. Nelson Pass based his testing / comments on the results he obtained with earlier Threshold amps and Polk "Cobra Cables", Matt Polk based his suggestion on testing the "Cobra Cables" with multiple different amps, etc... so you can see how they arrived at different figures. Polk shot for a "universal" Zobel and Pass had specific figures for his amps in mind. Jon Risch's suggestions are also somewhat "universal" but take into account some other important factors too. In this regard, he and i tend to think somewhat alike. I'll get to why in just a bit though.

Another person that comes into this equation is Bob Carver. Due to past experiences with "low inductance" speaker cables and some of his past amps, he builds "impedance compensation networks" or "Zobel's" right into the Sunfire amps. His thoughts are that he wants to keep the amp as stable as possible ( which requires a lower hinge frequency ) but at the same time, he doesn't want the Zobel's interferring with the treble response of the amp. As such, he's selected 80 KHz as the point where his high frequency protection kicks in at. This may also have to do with the fact that the Sunfire's use a high frequency power supply and this frequency also worked well to keep power supply noise from being transmitted through the amp and out to the speakers.

As to the way that i like to do Zobel's, i take several factors into account. That is, the nominal impedance of the speaker cable being used, the nominal impedance of the speakers being used and the bandwidth of the amp being used. You can start by selecting a frequency that you want the Zobel's to come into play. Personally, i like to keep them above at least 100 KHz. On the other hand, i think that the Goertz Zobel's come in at about 150 KHz ( give or take ), which is still fine for most amps and would be even less intrusive sonically.

Now you have to look at the nominal impedance of the speaker cable and the speakers being used. Let's say that we have a cable that is 2.5 ohms ( as Goertz is rated ) and speakers that have a nominal impedance of 4 ohms. Only thing is, when we get WAY above the audio band, that 4 ohm impedance is going to be MUCH higher. As such, the Zobel is actually running in parallel with the higher value impedance that the speaker presents.

In this specific case, i would use something along the lines of a 5 or 6 ohm resistor. When you place one resistor ( the speaker WAY above the audio range ) in parallel with another resistor ( the Zobel ), you automatically get a lower impedance because we are splitting the signal / sharing the load. Kind of like wiring two 8 ohm speakers in parallel and getting a 4 ohm load for the amp. In this case, our 5 or 6 ohm resistor is in parallel with what is probably dozens of ohms, so the impedance doesn't drop all that much. What it does do is present the amp with something that should be close to what the nominal impedance of the speaker is in the audio band. It also keeps the nominal impedance slightly above that of the speaker cable itself, acting as somewhat of a "meeting point" between the two. In effect, i've created somewhat of an impedance transformer over a specific frequency range. By maintaining a relatively consistent impedance both in and above the audio band, the amp remains more stable under dynamic conditions and high frequency transients / harmonic overtones aren't "stifled". This would normally occur due to what would be a much higher impedance load being seen by the amp, which would result in less power transfer and more ringing due to the impedance mismatch.

As Audioholics noted in their testing, when they terminated the speaker cable with the Goertz Zobel's by themselves, there was an impedance mismatch and some visible high frequency ringing / oscillation WAY up high in the MHz range. According to them, Goertz used a 10 ohm load and they measured the cables as having a nominal 8 ohm impedance. As i've mentioned before, terminating a line with anything other than the same impedance results in reduced transient response, increased ringing, reduced power transfer, etc... You could see the increased ringing in their tests as a result of the impedance mismatch.

When they terminated the Goertz with an 8 ohm Zobel, which is what they measured the nominal impedance of the cable as being, the ringing was gone and the cables were perfectly in phase with the output of the amp. Even though we are only talking about TWO OHMS of difference in terms of impedance matching here, the results were quite obvious. Now can you imagine how much POORER audio gear performs when you've got a 50 ohm preamp loading into a 75 ohm interconnect feeding into a 10,000 ohm amplifier???

All of the Audioholics testing follows my previously posted theories, right? So far, it all looks good on paper, right? One problem here though. What did Audioholics forget?

They forgot that the Zobel is connected in parallel with the speaker load in a real world circuit. As such, Goertz' 10 ohm Zobel would have actually looked like a 7 - 9 ohm load ( depending on the high frequency impedance of the speakers being used ) under normal operating conditions. As such, the Goertz Zobel would actually perform much closer to their "perfect" modeled 8 ohm results than what they show on their charts. Adding the extra paralleled impedance of the speaker across their 8 ohm Zobel would actually produce a lower impedance with slightly different (probably more than good enough though ) results.

Did you folks get all of this??? I know that some of it is kind of technical, but the more that you can learn, the less likely you are to be ripped off / talked into buying snake oil : ) Sean
>

Tommy: As one goes higher in frequency, the signal tends to travel more towards the surface of the conductor. Consequently, as one goes lower in frequency, the signal tends to travel through a deeper cross-section of the conductor. By making the conductor very thin yet maintaining a very wide & flat surface, all of the signal is conducted evenly regardless of frequency. This reduces time smear and maintains a more consistent series resistance / impedance regardless of frequency. This is yet another factor as to why you could hear increased "liquidity" with improved harmonic structure and timing of the notes. That is, each note / frequency has a very similar electrical path, length, series resistance and amount of surface area to travel. We'll call this "equal rights for all frequencies" : )

While one can obtain excellent results as far as skin effect goes with very small gauge round conductors, the problem is that the smaller gauge increases series resistance. In order to get around this problem, now we have to run multiple conductors in parallel. We now run into the problem of which geometry to configure these conductors in, how do we maintain the same spacing / EM fields between the conductors of the same polarity and how do we configure the two different polarities using multiple different conductors and how do we keep all the conductors of the same exact length? As you can see, Goertz' solution is a very simple yet elegant solution to all of those questions. That is, they followed the old "KISS" rule ( Keep It Simple, Stupid ).

As far as Magnan goes, if you check in the archives, you'll find that i've made some positive comments about some of their interconnects. I've never used their speaker cables or their "conductive paint" interconnects, nor do i think i ever will. I wouldn't mind trying out their speaker cables though, but i sincerely doubt that i would run it "side by side" as they suggest. This increases the inductance, which reduces the bandwidth and creates more phase errors. Like i said, the Goertz flat speaker cable design is simple yet elegant and solves all of those problems. Sean
>