Townsend Isolda cable

[restorer-john]

Although, the new 50 GHz TDR we got at work would probably show all sorts of wiggles into a speaker cable.

Take it home Don and have some fun. We'd all like to see your investigations.

 

[PierreV]

Just another reason to go active folks... no 'debatable' speaker cable lengths messing up fidelity.

Indeed. I can't imagine people or companies dishonest enough to claim cables would make differences in such setups.
And, if they ever did, I do not doubt that golden-ears reviewers would immediately expose the scam. I can't wait

 

[SIY]

Indeed. I can't imagine people or companies dishonest enough to claim cables would make differences in such setups.
And, if they ever did, I do not doubt that golden-ears reviewers would immediately expose the scam. I can't wait

There are quite a few cable peddlers selling very short wires just for such application. And of course, the long interconnect now needed, but using miraculous technologies unknown to engineers who design medical imaging equipment or space probes.

I poked around the Townshend website a bit. It looks quite typical for the genre.

 

[Thomas savage]

So many solutions we need to make some problems up to cope... And make some coin.

 

[DonH56]

Take it home Don and have some fun. We'd all like to see your investigations.

The problem is that, as with many of the test instruments at work, I can sign it out but I'm too nervous to bring home a piece of equipment worth more than my house... Guess I could bring home a couple of cables, though, and gen up some adapters to see just how improved the sound can be with 50 GHz RF cables used for speaker wires. The sad part is those RF cables, which I always though insanely expensive at a few thousand to $10k USD or so, are cheaper than many speaker cables these days.

Wasn't it Mogami who developed speaker cables with 8-ohm characteristic impedance many years ago? 1970/1980 timeframe... Too bad speakers do not have a constant 8-ohm impedance. I actually piddled, at the behest of some friends, with matching networks to counter the speaker's impedance swings and provide a matched impedance for that cable. The matching network thoroughly corrupted the frequency response (etc.) since the speaker was designed to be driven by a regular old amplifier, e.g. a ~0-ohm voltage source, and not a matched 8-ohm source. The speakers were B&W 801, my Magnepan MG-IIIa, and Quad ESLs IIRC -- a cross section of designs. None of them responded well even with "perfect" matching. Obvious to me today, but I was in college and all things were fun to try, and I got to show my friends (and teachers) the results of my experiments.

 

[Max Townshend]

It is only important to match impedance to the LOAD, not the Source. The transmission is simplex, not duplex, where the source would need 8 ohms impedance, but you would loose half your power, for no reason it is NOT a two way communications system It is only important to match the load impedance.

@amirm As I don't have the Isolda cable I can't do any comparisons so I have this 'theory' that would be easy to put to the test.

Assuming the high capacitance (and low inductance) is the reason for the improvements (regardless of theory) and other manufacturers exist that have high capacitance cables it occurred to me that maybe the rather high capacitance effect alone could be easily tested.

I would like to see the following test:
Generic (low resistance) cable with similar length as the Isolda with on the cable input 1x 3.3uH inductor (should be air wound and plenty of wire thickness to protect your amps) on the hot end of the cable and on the output of the cable a capacitor with about the same value as the measured capacitance of the isolda - measured (or estimated) capacitance of the generic cable.
After all the Isolda cable is just a very long stretched capacitor in essence that connects the amp to the speaker.

Just another reason to go active folks... no 'debatable' speaker cable lengths messing up fidelity.

The theory behind this test is to see whether that high capacitance is the reason or the waveguide theory is valid after all despite the theory saying the wavelengths are too short for this.
I can't see DC being improved by a wave guide nor say 100Hz for that matter.
But what I can 'imagine' is that when using a bandwidth limited 10kHz squarewave (which the output of any amplifier is) still has a lot of ultrasonics with a lot of energy well up to 100kHz for instance. We are clearly 'looking' above 50kHz here so the amp used should at least reach this.

At the end of the cable such a signal will be greeted by a capacitance of say 27nF or so (10m cable) which will have an impedance of around 50Ohm which is lower than the 'about 200 Ohm'. When this is in parallel with a speaker it will be lower.
Even better .. use the same Zobel as Towshend uses (10 Ohm + 0.22uF which is about 16 Ohm around 100kHz) and it becomes around 8 Ohm there.
No reason to 'reflect' anything there any more.
The extra capacitance will effectively short higher harmonics of say 1MHz or higher when present.

Another experiment could be to not even use the 3.3uH inductor and extra capacitor but only used the Zobel at the end of the cable ?

I assume mr. Townshend already has done these experiments and got better results with his (quite different geometry) cable.
In that case geometry is what matters.
When the same effect is achieved with just the Zobel network or the extra inductor + capacitor + Zobel then every handy DIY-er or handy business man can make 2 'cable interface boxes' and make any cable 'directional'.

I find the results Townshend got 'intriguing' enough to be tested by a third party and made more public.
When indeed Townshend's theory is true (reflections and geometry) then there is reasonable doubt that this matters in speaker cables.

 

[Max Townshend]

Why not try it yourself, as we did in the early 80s, by paralleling up 6 x 50 ohm coax cables (as shown in one of my Geometry Matters videos.), to get the same results. Make some yourself and listen to it. It has the unmistakable sound of matched cables.
The slow buildup of the voltage step at the speaker is NOT due to RC low pass filtering. It is due to the time it takes for each reflection from the amplifier (Z=0 ohms) to be absorbed by the speaker, from the grossly miss-matched cable. It occurs in a series of steps and after 500 or a thousand of these, the voltage eventually stabilises. The scope trace looks like a low pass filter trace, but this is because of the 100 mHz band-with of the scope. Look at the simulations and this measurement (fig 17). Please do the test on the 50GHz scope.

The Elephant-In-The Room is the multiple reflections on EVERY transient., adding noise and high frequency "brightness" to the sound. The bass and mid are little affected, but in Hi Fi, the Devil is in the high frequency detail.

With regards matching, it is not very important to match in the bass or mid as you would need a few square feet of cable cross-section area to match at 20Hz, to get R low enough. But, above 500 Hz, it matters. If you look really closely on a 100Mhz scope, you can see the ripples in the trace DNM 8R 0.05us attached. Most speakers have a reasonably constant impedance at high frequencies.

 

[Max Townshend]

Thanks for your reply.

The fact that a specific amplifier remains stable under simulations does not mean all amplifiers connected to certain speakers remain stable.

Also the 8 Ohm argument is nonsense as no speaker cable reaches anywhere near 8 Ohm at audio frequencies. Also there are no speakers that have an 8Ohm impedance >100kHz they may well be hundreds of Ohms or even lower.
Also amplifiers have a close to 0 Ohm output R and have to deliver current into complex (inductive/capacitive) loads and not in a nice resistive 8 Ohm load. So there is no impedance matching at all as the output R of the amp is not 8 Ohm.

Can you zoom in on those squarewave peaks to actually show they are reflections.

It is quite easy to see if they really 'distort' the signal by doing measurements and/or nulls at the begin and end of the cable (connected to a real load)

The second video seems to show resistive differences (given the low frequencies) and cannot possibly come from reflections. The wavelenghts simply are too long for audio frequencies.

 

[Max Townshend]

Man, that cable theory "white paper" is just begging for an annotated version.

For those of us who use cables for audio rather than shortwave transmitters, regular wire works just fine.

I am working on a full white paper on this. Just getting together the facts.

BTW why do you guys talk only about short wave transmitters when discussing impedance matching? What about the 50/60Hz grid? It would simply not work if impedance matching was not considered in the design. (I know a bit, because I have an honours-grade pass in low frequency transmission line theory)

 

[solderdude]

You seem to be repeating yourself. All of the above is already posted.

The part that makes me doubt about the audibility of it all is that transients in music never reach the risetimes of the squarewaves used for testing.

Can you show the reflections to exist using say... a 10kHz sinewave ? or even a 50kHz sine wave ?
Why would risetimes faster than 50kHz cause very audible differences when they can't even be reproduced ?

When reflections are bad in speaker cables due to the missmatch why don't they show up in interlink null tests.
Surely a few Ohm source resistance, a 50 to 100 Ohm interlink and 10k load would present a severe impedance mismatch ?
Why do we hear nothing when a few meter cable is nulled like we do on the speaker cables ?

Does it have anything to do with resistance and high currents ?

 

[Max Townshend]

From the same paper:

View attachment 23983

Exactly what instruments have you used to make this statement? My Audio Precision analyzer has a bandwidth of 1 Mhz. It can generate signals up to 200 kHz. It has a dynamic range better than any power amplifier by far. Its frequency response is ruler flat. In every way the analyzer exceeds our hearing sensitivity sometimes by massive (orders of magnitude) more sensitivity.

This article was written in 1998.

 

[Max Townshend]

Then we have this:

View attachment 23984

Spreading the myth that we need impedance matching for audio. We absolutely do not. A power amplifier has impedance very close to zero. Let's assume zero for simplicity. Let's have an 8 ohm speaker. Now, add a speaker wire with 8 ohm resistance/impedance and now you have a voltage divider:

Z1 will be your speaker wire and Z2 will be the speaker. Vin will be the amplifier. If Z1 is 8 ohm as Z2, then it will dissipate half the power the amplifier delivers to the speaker! Same current goes through both resistors so the math dictages power through Z1 is Z1 * I as will Power through Z2= I * Z2.

Raise your hand if you want your speaker wire to throw away half of your power. Good, I don't see any hands raised.

For RF situations or where cables are kilometers long as is with analog telephone we have reflections to worry about so matching impedances are used as you indicate. But for audio, we do not and as such, this is absolutely wrong thing to do. Wire resistance/impedance must be kept to minimum in order to reduce its losses.

In addition to straight losses, speaker impedance varies with frequency. If the wire resistance/impedance is a significant fraction of the speaker impedance, it will server to change its frequency response which again, is a bad thing.

I suggest reading Snow's 1957 Journal of AES paper, Impedance -- Matched or Optimum? Here is the synopsis:
View attachment 23987

View attachment 23988

This dog does not hunt. It didn't in 1957, and doesn't today. Simple, basic understanding of electrical systems and transmission stipulates this.

We have investigated this with forensic intensity and have found otherwise.

 

[Max Townshend]

Max, very nice to see you posting here , welcome to ASR,
Thank you for taking the time to personally respond to my query.
Keith

A pleasure. I look forward to the debate.

 

[Max Townshend]

The problem is that, as with many of the test instruments at work, I can sign it out but I'm too nervous to bring home a piece of equipment worth more than my house... Guess I could bring home a couple of cables, though, and gen up some adapters to see just how improved the sound can be with 50 GHz RF cables used for speaker wires. The sad part is those RF cables, which I always though insanely expensive at a few thousand to $10k USD or so, are cheaper than many speaker cables these days.

Wasn't it Mogami who developed speaker cables with 8-ohm characteristic impedance many years ago? 1970/1980 timeframe... Too bad speakers do not have a constant 8-ohm impedance. I actually piddled, at the behest of some friends, with matching networks to counter the speaker's impedance swings and provide a matched impedance for that cable. The matching network thoroughly corrupted the frequency response (etc.) since the speaker was designed to be driven by a regular old amplifier, e.g. a ~0-ohm voltage source, and not a matched 8-ohm source. The speakers were B&W 801, my Magnepan MG-IIIa, and Quad ESLs IIRC -- a cross section of designs. None of them responded well even with "perfect" matching. Obvious to me today, but I was in college and all things were fun to try, and I got to show my friends (and teachers) the results of my experiments.

The matching must be in the bulk of the cable itself. Not lumped at the ends. Please take a closer look at the simulations.

 

[March Audio]

Characteristic impedance is irrelevant for the wavelengths involved. You can’t change physics even if your wallet is involved.

This.

And this

Square wave with wide bandwidth is not a proper test unless it is band limited to the audible frequencies.

And this

.. Too bad speakers do not have a constant 8-ohm impedance. I actually piddled, at the behest of some friends, with matching networks to counter the speaker's impedance swings and provide a matched impedance for that cable. The matching network thoroughly corrupted the frequency response (etc.) since the speaker was designed to be driven by a regular old amplifier, e.g. a ~0-ohm voltage source, and not a matched 8-ohm source. The speakers were B&W 801, my Magnepan MG-IIIa, and Quad ESLs IIRC -- a cross section of designs. None of them responded well even with "perfect" matching. Obvious to me today, but I was in college and all things were fun to try, and I got to show my friends (and teachers) the results of my experiments.

 

[Max Townshend]

The problem is that, as with many of the test instruments at work, I can sign it out but I'm too nervous to bring home a piece of equipment worth more than my house... Guess I could bring home a couple of cables, though, and gen up some adapters to see just how improved the sound can be with 50 GHz RF cables used for speaker wires. The sad part is those RF cables, which I always though insanely expensive at a few thousand to $10k USD or so, are cheaper than many speaker cables these days.

Wasn't it Mogami who developed speaker cables with 8-ohm characteristic impedance many years ago? 1970/1980 timeframe... Too bad speakers do not have a constant 8-ohm impedance. I actually piddled, at the behest of some friends, with matching networks to counter the speaker's impedance swings and provide a matched impedance for that cable. The matching network thoroughly corrupted the frequency response (etc.) since the speaker was designed to be driven by a regular old amplifier, e.g. a ~0-ohm voltage source, and not a matched 8-ohm source. The speakers were B&W 801, my Magnepan MG-IIIa, and Quad ESLs IIRC -- a cross section of designs. None of them responded well even with "perfect" matching. Obvious to me today, but I was in college and all things were fun to try, and I got to show my friends (and teachers) the results of my experiments.

 

[Soniclife]

A pleasure. I look forward to the debate.

Max, these effects your referring to, do you have any measurements of the changes coming out of a speaker?

 

[Max Townshend]

Here is the latest experiment. The circuit is attached.

And here is a video.

https://drive.google.com/open?id=1HRRaiVVifHlSKF0A6B0ukfPLy4Enbc26

The Pink noise is from my phone on Audio Tool
The audio is from the Sheffield Lab test disc, only playing the left channel.
The amp driving the cables is a Cambridge amplifier
The second amplifier is a Nobsound Lindsey-Hood class A (an absolute audiophile bargain)
The speaker is a small two-way.
The spectrum display is from a calibrated microphone, 2 inches (50mm) from the middle of the speaker. The analyser is a Behringer Ultracurve Pro DEQ2496 .
The camera is a Cannon G12. The sound you hear, is from the microphone in the camera, positioned about 3 feet away from the speaker and is a little "shrill" sounding due to the appalling acoustics. But you can still hear the difference, even on your phone!
The dummy speaker, is a hard-wired implementation of the dummy load, as represented in the simulations. (a real speaker would be very loud and swamp the other)

The circuit allows input selection to the power amplifier and the two cables under investigation are in parallel, connected to the output terminals.
The two way switch on the right selects the cable.
The object of the experiment is to observe the difference between the voltage on the send end and the receive end of the cable, by amplifying the difference between each end of the ground ("-") wire.
There is a clear difference between the cables, with the Monster exhibiting rising output, commencing at about 500Hz, rising by about 12dB+ at 20khz. In contrast, the Isolda produces a "rolled-off" sound. Further, the Monster sounds a fair bit louder than the Isolda, a bit masked by the AGC in the Canon microphone. If the differences were due to DC resistance, then the Monster should be 75% SOFTER than the Isolda.

In the case of the Isolda, the difference sound is due to the resistance loss only, whereas with the Monster, there is a resistance loss, similar to the Isolda, but with a huge amount of added high frequency noise due to the decay of multiple reflections.

The mathematics, the simulations, the measurements and the listening all concur.

In forty years of research on the subject, I have no other explanation.

Please try the experiment yourself.

I am pleased to answer any questions.

 

[SIY]

I am working on a full white paper on this. Just getting together the facts.

BTW why do you guys talk only about short wave transmitters when discussing impedance matching? What about the 50/60Hz grid? It would simply not work if impedance matching was not considered in the design. (I know a bit, because I have an honours-grade pass in low frequency transmission line theory)

You do understand about wavelength, right? This is 5 or 6 meters of wire.

 

[DonH56]

Interesting factoid: Those loops of wire wrapped around some overhead power transmission lines are in fact there to help provide impedance matching, of a sort... They help get current and voltage in the proper phase to reduce power loss in the cables. Even at 50/60 Hz you have to do that when you have many miles of wire to contend with. At least I remember something from my decades-old power electronics classes...

I have a little article on transmission line effects in speaker cables on WBF and/or here if Amir brought it over. See e.g. https://www.whatsbestforum.com/threads/rf-speaker-cables.3759/#post-59881 (did not see it here). Since the rise time of a 20 kHz signal is about 17.5 us, reflections would have to be severe and sustained to be audible. In most systems the reflection at the speaker is absorbed, or mostly so, by the low source impedance of the power amplifier so triple-transit issues are inaudible. Assuming there is enough energy to generate a significant reflection to begin with, something I have not attempted to analyze. For audio circuits, reflections are usually buried in the rise time of the signal so do not appear as isolated events; the limited bandwidth of the system kills the stairsteps (bounce diagrams, anyone?)

EDIT: article is posted here now: https://www.audiosciencereview.com/...-analysis-of-speaker-cables-reflections.7154/