Not a real issue. Why would any thinking audiophile use a high capcitance cable in the first place?
Would you change your amp selection knowing...?
OK - so this thread was promted by some comments on another thread - not wanting to hijack that thread I created this one...
ISSUE: some high current designed amps have an issue with speaker cables that have a high capacitance.
- the amp can be driven to self destruction because of internal oscilation caused by the high capacitance of the speaker cable
- this does NOT apply to Tube amps - i.e. to my knowledge
The amps I know of that are affected in this way are Ayre, Gryphon and NAIM
- only NAIM warns of this up front AND instruct their dealers to let customers know about it
So why don’t other brands warn about the possibility?
QUESTION:
- would it put you off?
- would you select a different amp if the manufacturer warned of this "issue" up front?
Cheers
ISSUE: some high current designed amps have an issue with speaker cables that have a high capacitance.
- the amp can be driven to self destruction because of internal oscilation caused by the high capacitance of the speaker cable
- this does NOT apply to Tube amps - i.e. to my knowledge
The amps I know of that are affected in this way are Ayre, Gryphon and NAIM
- only NAIM warns of this up front AND instruct their dealers to let customers know about it
So why don’t other brands warn about the possibility?
QUESTION:
- would it put you off?
- would you select a different amp if the manufacturer warned of this "issue" up front?
Cheers
62 responses Add your response
Thinking of Wonder Cable, yes, this is possible. It warms up the sound, and enhances imaging. The oscillation part though, IMHO this is poor design. First, I don't know of any cable that has more capacitance than an ESL speaker. Second, it is easy to deal with this, and usually is, with an appropriate output snubbing network, or massively over-built output stage like Sander's Sound does. |
Hi Steve (Williewonka), I suspect that **many** high end solid state amplifiers are prone to adverse effects from speaker cables having ultra-high capacitance, at least if a Zobel network is not used with the cable. (My understanding, btw, is that if requested Goertz will supply such networks for use with their ultra-high capacitance cables). In extreme cases those effects may include destructive oscillations, as you indicated and as several members here have reported experiencing. In less extreme cases, though, there may be subtle but significant adverse effects on sonics, including things like overshoot, ringing, or low level ultra-sonic or RF oscillations that are not directly perceivable as such. Whether or not such effects occur will depend in part on how much feedback the amp uses, and on its gain, bandwidth, and output impedance. I’m somewhat surprised to see Ayre on your list of susceptible amps, btw, since most or all of their amps are zero feedback designs. In general I would expect lack of feedback to minimize or eliminate such sensitivity. To answer your question, though, the possibility of this issue, whether mentioned by the manufacturer or not, would have no influence whatsoever on my selection of an amplifier. It would certainly influence my selection of a cable, though. A cable having extreme and /or unusual parameters would be a non-starter for me. And if as is often the case basic parameters such as inductance, capacitance, and resistance are not specified for a particular cable, and if the manufacturer can’t or won’t supply that information, and if a rough idea of these parameters can’t be inferred from the cable’s description, I would look elsewhere. Best regards, -- Al |
Oscillation can self destroy the amp and might damage tweeters. I assume that amplifier can safely drive all typical cables and speakers unless manufacturer specifies limitations - like maximum load capacitance. Since preventing oscillation usually reduces bandwidth of the amp, either thru Zobel network, shallower negative feedback, limiting bandwidth at the input or frequency compensation, then amp that oscillates was likely designed with shortcuts to improve specifications (or poorly designed). |
@almarg - you bring up anothre very interesting point... In less extreme cases, though, there may be subtle but significant adverse effects on sonics, including things like overshoot, ringing, or low level ultra-sonic or RF oscillations that are not directly perceivable as suchSo another question one could ask - are members aware their amp could be suffering a less severe effect that may be degrading audio quality due to the speaker cables they have selected? We hear so many times people finding a particular speaker cable sounds nicer with their amp - was the old cable causing issues within the amp? @newbee ... How come cable manufacturers don’t warn potential customers?Well - for starters ... - it does not apply to Tube amps (as far as I know) - it does not happen with every solid dstate amp - high current only - the length of speaker cable increases capacitance. - it might even depend on the added capacitance of the speakers. So from a cable perspective it’s not an exact science However - i do think the amp manufacturer should make customers aware - just like NAIM does. After spending 40 years as an audio enthusiast there still are so many variables that the customer is expected to consider - thank goodness for Audiogon :-) Because the NAIM dealer told me up front - it made no difference to my amp selection - but if the amp had self destructed - I might be taking a different view on this question. |
I would not willingly buy an amplifier which could have such a failure. Nor would I knowingly buy such speaker cables. If I already owned such an amplifier, I would certainly expect the manufacturer to make it perfectly clear that the amp cannot be used with certain cable designs. And i think an amplifier manufacturer, knowing such cables are being sold, IS the party responsible for failure to warn the consumer. The manufacturer acting like they do not have any responsibility is a legal cesspool. On the other hand the cable makers also bear a great deal of responsibility to warn buyer of the pitfalls with certain amplifiers. |
@kiganki "then amp that oscillates was likely designed with shortcuts to improve specifications (or poorly designed).". It is obvious you know nothing about Odyssey amps. They are more honest than most audio manufacturers. Visit them at one of the audio shows and you will be singing an entirely different tune. |
The Goertz cables are flat and the conductors are closely coupled so as to increase the capacitance. This is done to reduce the Characteristic Impedance of the cable. Their cables approach 4 and 8 ohms (depending on cable selection) and so reduce reflections in the cable depending on the impedance of the speaker. Some may suggest that reflections at audio frequencies are a non-issue, but measurements that I and a team I was associated with (using a Time Delay Reflectometer) suggest that if you can get the cable to have a low CI, it will be more revealing. So there can be a reason for high capacitance cables, and amp designers need to suck it up. The old Polk Audio wire was high capacitance and did mess with unstable amps of the era. But as at least one pointed out here, ESLs are capacitive as well and are often driven by amplifiers :) |
Kalali, that’s an excellent question, and I’ve wondered the same thing myself. While I’ve seen a number of reports over the years of solid state amplifiers self-destructing as a result of having to drive cables having ultra-high capacitance, I don’t think I’ve ever seen a report of an amp being damaged from having to drive an electrostatic speaker. But while I’m not sure how to explain that, if I were to hazard a guess I’m thinking it may be related to the presence of the step-up transformer that I believe is used at the input of nearly all ESLs. Perhaps the bandwidth limitations and/or other characteristics of the transformer cause the amp to see a load impedance that is much less capacitive at ultrasonic and RF frequencies than it is at audible frequencies, and in comparison with the impedance of a highly capacitive cable at ultrasonic and RF frequencies. And my suspicion is that the destructive oscillations which have been reported to result from the use of high capacitance cables typically occur at ultrasonic or RF frequencies, not at audible frequencies. Also, I believe that the few ESL designs which don’t have a step-up transformer at their input, such as some older Acoustat models, have a built-in amplifier to step up the input voltage. In those cases presumably the built-in amp provides a relatively non-capacitive input impedance. Best regards, -- Al |
^I’m not sure if they still do. When I got mine the RC networks were separate. I’ve since read that some have them directly integrated into the cables. I’m not sure, but I seem to recall that separate RC network replacements were available for $20, but I also seem to recall that some might have received them gratis. Some have even made their own. |
Adding a network to an amplifier to correct for a cable is a tailspin, just adding eq to eq. I’d like to see how Goertz calculates a Z of 4 or 8Ω at audio frequencies from their geometry. SS amplifier outputs are a tiny fraction of 8Ω which is what gives rise to large damping factors. Characteristic impedances are beneficial when the source and load impedances are matched. Almost no speaker is a flat 4 or 8Ω impedance, largely negating any supposed benefit. Typical impedance variations of 4:1 are common and 10:1 is not uncommon. See Cable Snake Oil Antidote Amplifier Output to see how amplifier output impedance can interact with cables. |
I’d like to see how Goertz calculates a Z of 4 or 8Ω at audio frequencies from their geometry.Hi Ian, The table near the bottom of the following page of their website indicates R, L, C, and Z for their various speaker cables: http://www.bridgeportmagnetics.com/contents/en-us/d62_MI_AG_Speaker_Cables.html As I’m sure you are aware, characteristic impedance can be calculated to a close approximation as the square root of (L/C), using those parameters on a per unit length basis and provided that conductor resistance per unit length and dielectric conductance per unit length are insignificant. The L and C values shown in the table appear to be consistent with the indicated characteristic impedances, which range from "~1.7" to "~4" ohms. Best regards, --Al |
I have several sets of cables at home, my favorite are high count twisted litz. These cables definitely qualify as high capacitance. This capacitance problem is also effected by the speaker crossover itself that is attached to the cable. So depending on crossover design and layout combined with cable, a problem with speaker cables capacitance can occur with amplifiers that have a lot of feedback and some high-feedback push-pull tube amps. A capacitive load can drive the feedback phase far enough to lead to oscillation, sometimes at ultrasonic frequencies. You might not hear it, but soon there's smoke coming out of your tweeter. |
@almarg I’ve never used (L/C)^0.5. I conceptualize cables as a (SERIES LR with PARALLEL C) x Length. First sum the LR impedances and then add the inverse of the LR sum to the inverse of the C impedance. Z=1/(1/(ZL+ZR)+1/ZC) * Length. It's actually more complicated because ½L is in each lead and R is in both leads with the cap between them Using the numbers on the link for Divinity, 4nH .98mΩ 1.5nF / ft, I come up with ~0.05Ω @ 1KHz. The impedance is impressively flat relative to a 2 wire standard, but nowhere near 4Ω. This impedance is in parallel with the amp and speaker. Since the value is so low relative to the speaker impedance, the impedance remains low well past the audio band and can cause some amplifiers problems, particularly if the speaker has a very low Z minima. A ’benefit’ of plain old speaker cable is its impedance is rising, thus preventing amp problems. The downside is the rising impedance, quadrupling in the region where the ear is most sensitive, is reacting negatively in terms of phase. |
@ieales Hi Ian, I’m afraid I have to question or disagree with several things in your analysis: 1)I’ll start with the least significant of the issues that I see. What length are you assuming in your calculation that resulted in 0.05 ohms at 1 kHz? Plugging the numbers for the particular cable into your methodology I find that the result at 1 kHz is almost completely dominated by resistance, with the result therefore being not much different than the cable’s resistance spec of 0.00098 ohms per foot (x2 conductors, presumably, although that isn’t made clear in the table). 2)Your equation "Z=1/(1/(ZL+ZR)+1/ZC) * Length" would reflect the parallel combination of (ZL + ZR) and (ZC), yet as you correctly state L and R are in series, while C is in parallel. 3)Related to that, specifically to the fact that L and R are in series, I don’t see the basis for your statement that "this impedance is in parallel with the amp and speaker." Certainly the amp is not being loaded with 0.05 ohms! 4)Most significantly, I believe you are conflating "impedance," derived as a combination of the individual impedances of R, L, and C at a given frequency, with "characteristic impedance," which is not the same thing. I recognize that the two terms are sometimes used interchangeably, but that is incorrect and potentially confusing. (Even the heading in the Goertz table that I referred to makes that mistake, although the writeup above the table makes clear that they are referring to characteristic impedance). For example, a 75 ohm coaxial cable has a "characteristic impedance" of 75 ohms, but at most frequencies certainly does not have a 75 ohm "impedance" based on any series and/or parallel combination of the individual impedances of R, L, and C at each frequency. "Characteristic impedance" is essentially independent of frequency, assuming, as I alluded to earlier, that conductor resistance per unit length and dielectric conductance per unit length would not affect a calculation based on the square root of (L/C) significantly. See the Wikipedia writeup on "Characteristic Impedance," which is consistent with my understanding of the matter. Best regards, -- Al |
@almarg Hi Al, good spotting. ~<|:-/ Actually, I've never seen the formula Z = ( L / C )^0.5. Where does it originate? 1) 25 foot length as that is what was shown on the MI/AG site in Fig 4. 3) I mistyped. Conceptually, the cable LR are in series and the C is parallel with the load. 4) I understand characteristic impedance. A 75Ω cable is designed to be driven by 75Ω source and terminate in 75Ω load impedances. I dealt with PCB impedances for years in high speed digital and have fixed innumerable CATV issues by changing splitters or terminating open jacks with 75Ω loads for friends and family. |
Hi Ian, Thanks for the clarifications. While Zo (denoting characteristic impedance) = ( L / C )^0.5 is an equation that is widely used in various EE applications, I don’t recall ever seeing a **simple** derivation of it. The Wikipedia writeup I referred to on Characteristic Impedance, in conjunction with the Telegrapher’s Equations writeup it links to, provides a derivation, although it is rather complex. Another derivation is shown at this link in the first answer to someone’s question. Note that in both derivations the bottom line equation which includes series resistance R and shunt conductance G reduces to Zo = ( L / C )^0.5 when R and G are zero, and therefore to a close approximation of that equation when R and G are small enough to be negligible (on the same per unit length basis that is used for L and C). And under those circumstances Zo becomes essentially independent of frequency. Keep in mind also, as you probably realize, that characteristic impedance is essentially independent of length. Best regards, -- Al |
I wonder if it's a matter of the math not matching the measurements, unrecognized variables, or errors of another sort? For your consideration: (sorry, but most of the links no longer seem to work.) The authors suggest that the supplied RC networks values for the MI 2's could be improved upon, perhaps someone might be able to confirm or refute this. Furthermore, I would be most appreciative(!) if instructions for construction of the ideal zobels for the MI 3's could be provided. |
| Post removed |
| Post removed |
^I'm not sure if my speakers impedance load is a consideration, but if it is: https://www.stereophile.com/content/thiel-cs35-loudspeaker-measurements |
A few Cardas cables can be pretty high... http://www.cardas.com/clear_sp.php This one is 278 pf / foot x 10 = 2780 pF for a 10 ft cable- that's pretty high This is probably their highest capacitance cable By comparison - The Van den Hul D352 is 32.5 pF / meter. so that's about 100 pF for a 10 ft cable Don't let that put you off Cardas - some of their cables are much lower At least Cardas makes the capacitance of their cables knowm - which makes me think they are aware of the issue - and feel the audiophile should know about it Many companies do not Regards - Steve |
Speaker Cable:Hello Steve - I copied the above from the Blue Jeans site. I wonder if their, "...barring a really odd design, which may introduce various undesirable effects...." is tacit acknowledgement of the oscillation risk you raise (among other things, I suppose). In reply to your original question, knowing an amp was susceptible to high capacitance induced oscillation would not put me off from buying that amp (assuming it was desirable to begin with) BUT I would certainly exercise care with the speaker cables attached to it. Thanks for the Cardas info. By contrast to the high capacitance number for their Clear spkr. cable I note their Parsec cable is 30 pf/ft. If bi-wiring, how would these capacitance numbers work...simply additive? or something else?? Anyone have an idea about capacitance of the Clear Day Double Shotgun speaker cable? Thanks again. |
If bi-wiring, how would these capacitance numbers work...simply additive? or something else??Parallel doubles capacitance and halves inductance and resistance. Bi-wiring separates the load into two parts. Response for each part differs from a single drive. Don't know a thing about Clear Day and they provide no technical information. From reading their information, it seems their product comes from 'messing about' and may be very specific or just generic. Solid core will be stiff. Price seems too low for pure silver of a reasonable gauge, say 18. That's about 4oz of pure silver ~$65. 18ga solid silver hookup wire is about $9/ft. 32 feet for two 8' runs is ~$288. Add in connectors, heat shrink and labor and the price is too low. |
ieales - Thanks for the bi-wire capacitance info. Don’t want to sidetrack this thread into a defense of Paul Laudati and Clear Day Cable though I will say I’m inclined to take him at his word. Paul has been around a while and it would be a fairly easy task to "autopsy" one of his cables and refute his solid silver claim if it were bogus. To my knowledge, that has not happened. I can report the Double Shotguns were better with my gear than Morrow Audio SP-4 silver plated copper...and not by a little bit. They are also preferred in one amp/speaker combination I have over Cardas Parsec cable. The review at the link below, while largely anecdotal, might be of interest. http://www.highfidelityreview.com/clear-day-audio-cables.html |
