Expensive Cables: At AXPONA earlier this year, I was in one of the many listening rooms where there was a cable demo. As an engineer, I've always been skeptical if there are no measurements or technical explanations that make sense. But, in this demo, they played 1 minute of a song with male, female vocals, instruments, and some percussion. Good variety for a minute. Then they replaced (get this) the POWER cable to the CD player. Played the same music again. After some discussion, this was repeated with cable #3 and finally, cable #4. Now, at 72, (no I was only 71 at that time) my hearing is not quite out to 22kHz as it once was, but I could actually hear a difference between cables 1 and 2 and again between 3 and 4. Many others in the audience heard a big difference between 2 and 3 where I heard no difference.
The next day I returned and took a seat at front row center. Again, I did hear a difference. If this is actually true, it overturns everything I know about good engineering and that is a scary thing.
They claimed that the most significant improvement is made by upgrading power cable to the source as the first upgrade.
I'll be back there next year to listen again.
I also contend that I could have 2 identical amplifiers, one in a very well designed case, the other in a contact paper covered cardboard box and the audience would always pick the one in the good case even if the 2 identical amplifiers were swapped.
The next day I returned and took a seat at front row center. Again, I did hear a difference. If this is actually true, it overturns everything I know about good engineering and that is a scary thing.
They claimed that the most significant improvement is made by upgrading power cable to the source as the first upgrade.
I'll be back there next year to listen again.
I also contend that I could have 2 identical amplifiers, one in a very well designed case, the other in a contact paper covered cardboard box and the audience would always pick the one in the good case even if the 2 identical amplifiers were swapped.
Hi Dan,
I reasoned that the reactance of 0.01uF is about 100 ohms at 159 kHz, and that at around that frequency is where things get important, such as the cable transmission line effects and is also at least an octave or two below the gain crossover frequency of most audio power amps. At the same time, I did not want to make it bigger than necessary, since that could increase dissipation under some conditions. I think when all things are considered, the overall optimum is quite broad. So, for example, one would likely be perfectly fine in using 0.1 uF, since under any reasonable listening conditions there is not that much sustained energy at the higher frequencies.
I have often wondered about why speaker manufacturers do not routinely put such a Zobel network inside their speakers. It is probably just an issue of attention to detail, and that many speaker designers are not as familiar with power amplifier behaviors. Of course, for all I know, there may be some speaker manufacturers that do put a network in and just don't advertise it
.Cheers,
Bob
Was it Nordost?
I have also often been skeptical as well. Probably a little more skeptical of power cords than speaker cables. If the role of the power cord is to suppress RFI/EMI ingress, I would think that a good component would already have adequate line filtering, but maybe they don't. Apart from reduction of RFI/EMI ingress, it is hard for me to understand how the last 6 feet of line voltage delivery to the unit could make much difference, when it is usually preceded by up to 50 feet of 12 or 14 AWG Romex. Maybe we need a Zobel network on the line
. Cheers,
Bob
Ok thanks Bob, RNMarsh reckons 100kHz is into TL behaviour for 100R/Z Cat5 pairs, agreed most fig-8 twin speaker cables will be around 100R.
There are speakers with impedance eq to say 7R across audio band but no mention of US/RF (say 50kHz and up) input impedance.
I have not seen graphs of real speakers up to say 200kHz, and I have never seen cable RF termination inside a speaker box.
The outstanding majority of mass market and hi-fi speakers I have serviced run 12dB filters using inductive series bipolar electros (tweeter), no compensation networks and internal wiring and cabling with significant loop areas, IOW totally undefined HF impedance characteristic.
In practice and by experience I would say such matched dumping/damping of unwanted US/RF energy is mandatory.
There are speakers with impedance eq to say 7R across audio band but no mention of US/RF (say 50kHz and up) input impedance.
I have not seen graphs of real speakers up to say 200kHz, and I have never seen cable RF termination inside a speaker box.
The outstanding majority of mass market and hi-fi speakers I have serviced run 12dB filters using inductive series bipolar electros (tweeter), no compensation networks and internal wiring and cabling with significant loop areas, IOW totally undefined HF impedance characteristic.
In practice and by experience I would say such matched dumping/damping of unwanted US/RF energy is mandatory.
Last edited:
Plugs/sockets, coppers, insulations/pigments/fillers are different is the answer, for the experiment it's worth making an IEC AC lead using the same Romex that's in the walls to keep the 'continuity'.
I run white sheathed power cables exclusively, I have grey and black IEC standard power cables also and they all sound different for what are essentially identical cable assy's.
It's already on my to-do list, I have suitable parts from old high power smps/solar convertor boards waiting for me.Maybe we need a Zobel network on the line
What measurements do you want? The DCR can be calculated from the gauge. The RF impedance is 100 Ohms +/- 5 Ohms measured to 500 MHz (Cat 5 E) and higher for Cat 6 & 7. Most reputable cable has been tested and certified to UL standards on a network analyzer.
RE Zobel on cable- My concern was that an amp with a higher output Z may have a frequency dependent interaction with an RC network. And in production on a premium product the 100 Ohm caddock film resistors were not that expensive.
I have been using a Zobel (.1Uf and 100 Ohms) on power cords to deal with RFI stuff. At first glance it seems preposterous that it could make a difference, until you map out the network of signal cables and power cables and look at the RF antenna they can represent. Transformers etc. are not free of capacitance and leakage at both power line frequency (120 to 240 V does not need a lot of leakage to impose some current on a ground that will generate some uV of noise) and RF. RF can pass through and get rectified all over the place. Power line communications can push as much as 3A at up to 30 MHz across the line.
Nordost cables are really expensive and the hype is pretty unfathomable. I regard there stuff as a form of male jewelry. Decorative perhaps but light on real technology. They do have some totally entranced reviewers however.
Hi,
While I do understand use of zobel network for stability and conducted emission, I don't understand why it would have any major effect on emi, rf radiated injection. I haven't done any measurements and there might be somerthing I miss.
My assumption is that radiated injection would be common mode. Then perhaps a small capacitor at connector to ground/chassie would be better?
/örjan
If possible a Z sweep of Cat5 wired as speaker wire (the 4 pairs paralleled).
IOW I want to see at what frequency paralleled Cat 5 becomes TL, and the Z absolute value....please.
I agree you provided ideal solution (for 100R cable).
Yes, house wiring should be damped to quench RF pickup and damp noise due to household loads.....anybody have an idea of RF impedance of 'twin and earth' building wire ?.
I did not say that was Cyril's amp, all I said was it was a blameless with a 10K feedback
resistor, shunted by an RC with 330 ohms for the R which matches comments made by
Cyril in the write up. The cap value did not match - pure guessing game on what amp
Cyril used.
Edit: I went back to Cyril's article and further down he mentions that his amps used wire
wound .22 R emitter resistors and he measured them to have .3uH. He also mentions that
D. Self states that these should be non-inductive so he replaced them and states that the
oscillation was worse! I think he needed to look at the global loop compensation rather
than component swaps - he also changed bypass caps for a minor improvement.
Last edited:
For quite some time I have used 3-watt metal oxide film resistors for emitter resistors - usually 0.22 ohm - and they seem to work well and have adequate dissipation. They are naturally non-inductive and are easy to get and inexpensive. They are able to withstand high current for short periods of time without failing.
Cheers,
Bob
No, you've just discovered that expectation bias is a huge effect. The fact different people had different expectations leads to differences between people's response. You absolutely must use double-blind testing to measure psychoacoustic perception - even watching someone's lips changes your perception of sound, google "McGurk effect".
With double-blind testing expectation is no longer correllated with the equipment choice, so that only real audible effects and statistical noise are present in the results - take enough measurements and the signal emerges from the noise, as usual.
Yes, another excellent example of expectation bias. The mind's emotional response to sound is not fixed, its totally up for grabs by almost anything other than the sound itself! So many people have no idea this happens (its very different from visual perception, for instance). The McGurk effect is startling if you've never seen it before - what you see completely overrides auditory perception.
I guess its a question of orders of magnitude. Common speaker cable lengths are 3 to 5m. 5m is about 15ns, or a 1/4 wave at 20MHz, and many early amplifiers were essentially passive devices over 1MHz, so that resonances of the transmission line were damped and never likely to affect the amp - so long as it could drive 1nF or so. Any severe RF interference would affect the preamp inputs first (not such an issue with modern digital sources).
More modern amps have higher bandwidth devices, leading to a situation where the resonance of the leads as a transmission line starts to be able to interact with the amp.
This is an argument for using audio amps with limited bandwidth, which is in conflict with the use of more open-loop bandwidth to reduce HF distortion.
Its also an argument for paying more attention to speaker line termination at the amplifier - if RF can't pass between amp and speaker cable, in either direction, and the line is roughly terminated as 100 ohms for RF, no need to worry about the speaker end of the cable. Its pretty crucial that the feedback take-off is protected from RF on the speaker line, so must come before any series inductor or other such filter.
It strikes me class-D amps may be particularly sensitive given the feedback loop is there mainly to linearize the output (reconstruction) filter, and no separate RF blocking filter is normally used after the reconstruction filter, exposing the feedback to junk on the speaker cable.
Last edited:
Yes, house wiring should be damped to quench RF pickup and damp noise due to household loads.....anybody have an idea of RF impedance of 'twin and earth' building wire ?.
My guess would be:
Above 1 MHz they would be in 50 to 150 Ohm range. But each brand or type has a different construction and different RF impedance.(more likely above 100 Ohms) Then again most power cords & cables have huge losses above 1 MHz.
My guess would be:
Above 1 MHz they would be in 50 to 150 Ohm range. But each brand or type has a different construction and different RF impedance.(more likely above 100 Ohms) Then again most power cords & cables have huge losses above 1 MHz.
Last edited:
100 ohms or so for any two wires, so perhaps 50 between live and the earth/neutral as a combination. And yes PVC insulation is poor for RF propagation at VHF and up, big loss factor / moisture absorption.
Mains wiring also carries common-mode signals, ie acts as a set of antennas, and the size and extent then become important, not the impedance.
I suspect that terminating the speaker cable with 100 ohms at RF at the amplifier end might not be desirable or practical, depending on the frequency at which the termination begins to look like an adequately functional termination. Suppose, for example, that we want adequate termination to begin at 200 kHz, roughly the number we have talked about for use of a Zobel at the speaker end. Let's define adequate as where the reactive component of the termination is about the same as the resistive component. This was the case where the 0.01uF capacitor had reactance of 100 ohms at 159 kHz. If we use a series inductor in parallel with a resistor of 100 ohms to series-terminate the line in the amplifier, and we select the inductor value to be 100 ohms at 200 kHz, then at 20 kHz the reactance of the inductor will be 10 ohms, and this will really kill damping factor and frequency response at 20 kHz.
Of course, if we argue that the termination need not become an adequately functional termination until 10 MHz, for example, then things are not quite as bad. Also, a substantial inductance in parallel with a fairly large resistance of 100 ohms would result in large ringing when feeding a capacitive load. I know I have made a lot of assumptions here and hope that I did not mis-read what you were suggesting.
I think your comment about class D amplifier susceptibility to EMI ingress is a very good point.
Cheers,
Bob
A transmission line with a grossly mismatched far end cannot be parallel terminated at the
source end. The far end mismatch causes a nasty input impedance to the TL and 100 ohms
in parallel with a nasty impedance is still bad. In digital logic and video for example, you
can source terminate (series termination) a TL by having the source impedance match
the TL Z0 but this is not practical with an audio power amp.
source end. The far end mismatch causes a nasty input impedance to the TL and 100 ohms
in parallel with a nasty impedance is still bad. In digital logic and video for example, you
can source terminate (series termination) a TL by having the source impedance match
the TL Z0 but this is not practical with an audio power amp.
...hard for me to understand how the last 6 feet of line voltage delivery to the unit could make much difference, when it is usually preceded by up to 50 feet of 12 or 14 AWG Romex. Maybe we need a Zobel network on the line
Here I see 35' #12 to fusebox, then 550 feet of saggy #2 Al to street, a transformer, then several+ miles of 20kV 3/16" steel to the substation. Most of my neighbors have welders. The last 6 feet is a mere gnat in the eye.
The idea of the power company damping the power lines, other than operational efficiency, is absurd. Lights glow, motors spin. If I need "better" power, that's my problem. And stepping aside, I would object to PECo dressing up the lines for cleaner power at my expense (unless they did me first, and if I cared).
Cordell may be (generally IS) right. But it took a long session on the board to see why. 100r + 0.01..0.1u all seem fine, and "all things" must consider the speaker.
I started with a basic R-L model of speaker impedance over 5kHz. I got some mighty sharp resonances, around 160kHz. While Cordell's "audio" amps loaf at 160kHz, I remember when that was a real danger-zone for most transistor amps. While many Tigers died, a few survive to this day.
Ah! SPICE coil is perfect, real speaker inductance is real lossy. Eddy current in the pole. An LR network is a narrow approximation, a string of LR is better. The Dodd paper surveys several models against a real speaker, however it is a woofer. Applicable to Full-Range speakers but not boxes with tweeters. No values given either. I "know" the shape of a speaker's Z above the audio band and hacked the curve.
With eddy-coil loss the curves get smooth and the "optimum is broad". Anything from some less than 0.01u to say 0.1uFd is fine. (0.3u may be over-damped and also absorbs cymbal-power.)
Part of the reason for "no preference" is that different tweeters will have different inductances and losses. So the curves will vary. Enough to matter? My shoulder is sore, not today.
BUT: I assumed a normal voice-coil tweeter. There are electrostats. I have been touted a "panel" which clearly had a Motorola piezo (made trumpet sound like plastic). The Z of these goes DOWN above the audio band (there may be some stopper resistance). Then a zobel at the speaker does about nothing; of course some amps still object to naked capacitance.
> metal oxide film resistors ... <snip-snip> They are naturally non-inductive...
Let's underscore low value like 0.022r emitter resistors. Above some Kohms they may be spiral-cut to raise resistance. Dunno if that matters "at audio". With some transistors working to GHz it may matter inside the amp.
Attachments
I'm saying disconnect the amp from the line at RF, then terminate it at RF. The rest of the world sees a terminated transmission line at RF, sees the output of an audio amp through an inductor at lower frequencies. The amp can't see the bad impedances through the inductor as any normal speaker line cannot be a 1/4 wave stub below several MHz. Of course its all a bit messy at middle frequencies, and a more carefully engineered set of filters is more complex and this a good reason not to do it (unless you have to).
Some of the termination theory with speaker cable has been suggested a long time ago. The subjective results I got from applying some of these ideas were inconclusive.
I remember being given a complimentary pair of super cables to try out. I really did not want them and when a friend bought himself a new amplifier I gladly gave these to him.
This new amplifier which had endorsement from the audio press had to be marginally stable. With the super cable following switch on there was a loud hum, some heat and a whiff of burning followed by silence.
The amplifier was replaced and with the dealer present the experience was repeated at switch on. Fortunately another situation was averted by rapid turn off. The dealer confiscated the super cable.
I like the perspective taken in Jensen Application Note - 001 "Some Tips On Operational Amplifier Stability" Jensen Transformers Application Notes | Jensen Transformers which may be of interest to others.