George, you and Ralph are talking about two different things. You are referring to the Fletcher-Munson effect. Ralph is referring to the increase in output impedance of a passive preamp that occurs as the volume control POSITION is reduced from maximum, to achieve a NORMAL listening volume. That increase in output impedance increasing the effects of the cable that is being driven.
That said, it's not clear to me why high output impedance + high cable capacitance would affect bass frequencies. Certainly if both parameters are extreme enough upper treble response could be affected, due to the two parameters combining to produce an RC low pass filter effect. And if the input impedance of the destination component varies significantly with frequency, or is too low with respect to preamp output impedance, bass performance could conceivably be affected. But I'd be interested in further explanation of how high output impedance could interact with cable parameters to affect bass frequencies.
Regards,
-- Al |
01-13-11: Georgelofi
To give some measured examples of interconnects
1: With IC's at 100pf per foot a 1mt = approx 300pf
this with the Lightpeed at it's highest output impedance has a HF rollof of -3db at 76khz
2: Same with 200pf per foot 1mt = approx 600pf
this with the same gives HF rolloff of -3db at 38khz
3: At 300pf per foot same gives a HF rolloff of -3db at 25khz.
Good interconnects are usually below 200pf per foot. George, from those numbers I infer based on bandwidth = 1/2piRC that the maximum output impedance of the Lightspeed is around 7K (which I see is confirmed in your post immediately above). Other passive preamps may be considerably higher, however, at worst case settings. And the issue concerned compatibility with long interconnects, not 1m interconnects. 15K ohms, for instance, in combination with 15 feet of 50pf/ft cable results in a -3db bandwidth of only 14kHz. Ralph & George, thank you for the explanations concerning bass effects. Just to clarify, though, it should be noted that those effects are essentially unrelated to cable capacitance or cable length. Best regards, -- Al |
George, note that the issue is not whether or not the volume control will have to be used too close to the bottom of its range. The issue is that even with the volume control set to its maximum position, a substantial fraction of the power capability of the amp (perhaps half) will not be able to be utilized. And keep in mind that the gentleman's speakers are relatively insensitive, at 83db/1W/1m.
I doubt it will be a problem, as I indicated, but that is the issue.
Best regards,
-- Al
|
Paul, I'm not certain but I suspect that the reason for the Atma-Sphere S30's high sensitivity number may be that it reflects the voltage DIFFERENCE between the + and - signals at its balanced input, which would be twice the amplitude of either signal. Not sure how its rca input is configured, or if the 3V sensitivity spec is meant to apply to that input as well as to its balanced input.
Also, although I realize you were referring to a 92db speaker just in answer to Jult52's question about "why does it matter," it's worth noting that the MMG's sensitivity is spec'd at 86db/2.83 volts/1 meter, which for its 4 ohm nominal impedance is only 83db/1 watt/1 meter. Although as a planar speaker its acoustic output will fall off relatively slowly as listening distance increases.
Jult52, the only ways that occur to me for stepping up the dac's output voltage would be:
1)Putting an active gain stage somewhere in the path between the dac and the power amp, but of course that would defeat your purpose in using a passive preamp.
2)Modifying the dac internally -- lots of potential issues there.
3)Using some sort of audio step-up transformer -- lots of issues there as well.
Best regards,
-- Al |
Re batteries vs. wall-warts, perhaps the reported sonic differences are due to voltage differences? I realize that the Lightspeed contains an internal 5V regulator, but perhaps differences in the "12V" input voltage result in different internal operating temperatures, perhaps affecting sonics in some way.
Also, as a point of information, lead-acid batteries (such as the one Anthony linked to), exhibit much greater changes in output voltage as a function of charge state than lithium-ion batteries do.
Best regards,
-- Al |
Good point. The prongs which can be seen peeking out in the photo of the charger kind of look like it's probably USA-compatible. And the three adapter plugs it pictures further down the page are obviously not for the USA, which would seem to imply that the built-in plug is USA-compatible. George, can you confirm that?
Also, a number of Hong Kong sellers offer similar battery assemblies on eBay, indicating that they are supplied with whatever plug type corresponds to the buyer's country. Search under "12V Li-ion Rechargeable Battery CCTV."
Best regards,
-- Al |
Is there a Lithium battery supply you would recommend? I don't see any reason why the one George linked to above wouldn't be a good choice. Best regards, -- Al |
Hi Paul,
Yes, there are significant differences.
An active buffer stage will provide a gain very close to 1, while providing a high input impedance (to the stage itself) and a low output impedance.
A transformer or autoformer will transform impedances in proportion to the square of the turns ratio that is selected, while at the same time causing a voltage reduction (or gain) proportional to that turns ratio. Example:
If a TVC is set to provide 12db of attenuation, that corresponds to providing an output voltage that is 1/4 of the input voltage. So the turns ratio for that setting would be 0.25. The output impedance would be ((0.25)squared) or 1/16th of the output impedance of the source component. The load impedance seen by the source component would be 16 times the input impedance of the destination component (presumably the amplifier).
If the TVC were set for unity gain (no attenuation), there would be no impedance transformation, the output impedance would equal that of the source component, and the input impedance would equal that of the destination component. But as the amount of attenuation is increased, the output impedance will decrease rapidly, since it is proportional to the turns ratio squared.
Best regards,
-- Al |
As I see it the 10:1 rule will be of no importance if all of the following conditions are satisfied:
1)The output and input impedances that are involved are both essentially resistive, and do not vary significantly as a function of frequency. That condition will usually not be satisfied if the preamp has a coupling capacitor in series with its output, as is the case with most tube preamps and some ss preamps. I'm uncertain as to how often it will or won't be satisfied if no capacitor is present.
2)The component supplying the signal can handle the load impedance without distortion or other signs of stress. I would not expect that to be a problem in most cases unless the load impedance is very low (e.g., 1K or less).
3)The loss in gain that will result is not an issue. For instance, a 1:1 ratio would result in a 6db loss in the maximum volume that can be achieved, in comparison to a high ratio.
Also, cable capacitance must not be too high in relation to output impedance, but that is a separate issue.
BTW, if the Air Tight's 600 ohm output impedance specification does not make any reference to frequency, and if the design includes a coupling capacitor at its output, the output impedance is likely to rise to considerably greater values at very low frequencies (perhaps 3K or 4K at 20Hz). With a 15K load impedance, that would result in a small but very conceivably audible bottom octave rolloff. A nominal (presumably mid-frequency) output impedance number should be multiplied by 50 to 75x, rather than 10x, to be completely safe under all circumstances, IMO. The 10x number would be appropriate to use if the worst case output impedance is known.
Regards,
-- Al |
That has to be the lowest capacitance if any cable - no? The extremely low capacitance results from the flat construction with wide spacing of the conductors. 300 ohm antenna twinlead has similarly low capacitance. The lack of shielding could be a problem in some setups, though. Also, the relatively high resistance and inductance of the return conductors (compared to what a shield would provide) might also contribute to ground loop issues in some systems. Best regards, -- Al |
05-31-11: Georgelofi
... Keep interconnects as short as possible (even if it means reconfiguring the system) and to get ones that have low capacitance less than 100pf per foot. George, is that what you really meant to say? I would not consider interconnect cable capacitance that approached 100pf/ft to be anywhere near being "low." For instance, just 6 feet of 100pf/ft cable would have a capacitive reactance at 20kHz of about 13.3K, certainly low enough to be significant in relation to the output impedance of your LSA, which is around 7K iirc. Best regards, -- Al |
Hi Paul,
Yes, I certainly don't question that, assuming there are no significant tradeoffs in the "no contact" approach. As I indicated, what I was questioning was simply the explanation that I quoted.
Best regards,
-- Al |
09-22-11: Clio09
In my conversations with some designers the indication was that active preamps are responsible for a larger source of the noise in ones system than amps (assumption is we are talking similar designed equipment).... Adding any passive preamp I own (a TVC and the LSA) to the chain and then shorting the inputs on the preamp the noise level does not change (regardless of volume level). I can't say that for active tubed preamps (mine and other's systems), especially those that use tubes. Interesting, Anthony. I would add the thought, though, that the increase in noise that occurs with an active preamp could often be due in part to ground loop effects occurring between the active preamp and the amp, especially if the interface is unbalanced. To the extent that may be the case, a dilemma would arise, both philosophically and technically, as to which component is really responsible. Paul, thanks for your comment. Best regards, -- Al |
Agreed, Anthony. That certainly reinforces the meaningfulness of the experiments you cited. This Jensen paper may be of interest to others who are following this thread. It explains why unbalanced interfaces between components can be prone to both low frequency hum and high frequency buzz, resulting from ground loop effects. Best regards, -- Al |
Gentlemen,
In a situation where the LSA is being driven by a component whose output impedance is low to the point of insignificance at all frequencies (which appears to be the case here, with the Metrum Octave DAC being solid state and having a nominal output impedance of 85 ohms), and where the LSA is driving a load impedance that is purely resistive (i.e., the Rothwells), and *IF* the LSA truly acts at any given volume setting like a pair of resistors, I don't see why it shouldn't be able to drive 30K without issues.
IIRC the LSA's output impedance may be as high as about 7K when driven by a low impedance source. Obviously 30K/7K falls far short of meeting the 10x guideline, but if all of the impedances that are involved are essentially resistive, and therefore constant as a function of frequency, why would there be a problem (other than the inability to drive the amplifier to full power, as I had pointed out)?
Best regards,
-- Al |
Devilboy, I have a pair of the 10db Rothwell's, and as measured with my not particularly accurate analog multimeter the series resistor has a value of about 21K, and the shunt resistor has a value of about 9.5K. As you realize and as George has confirmed, the 470K input impedance of your amplifier certainly presents no issues. (As you no doubt realize, the attenuators should be placed directly at the input connectors of the amp, so that there will be no interaction between their output impedance and cable capacitance).
With the attenuators so located, and driven by the LSA via a short cable, the LSA would see a load impedance of about 30K. I suspect that is not an issue either, based on the assumption that the impedance "looking back" into the LSA is essentially resistive and does not vary significantly as a function of frequency, given that you will be driving it with a solid state DAC having low output impedance. Hopefully George will confirm that assumption.
You will not be able to drive your amp to full power, though, even with the LSA turned up all the way, as 10db attenuation will reduce the DAC's 2V maximum output to about 0.63V.
For most volume control settings of the LSA, btw, the attenuation resulting from insertion of the "10db" Rothwell's will be more like 11 or 12db, because the impedance "looking back" into the LSA will sum together with the Rothwell's series resistor.
Concerning the sonic effects the Rothwell's may have under properly impedance matched conditions, I'm not using them with the VAC Renaissance 70/70 MkIII amplifier I currently have, but I noticed no adverse effects when I previously used them with a lesser quality EL34-based Paxthon amplifier.
Regards,
-- Al |
10-11-11: Devilboy
Jeez, if 470K isn't enough, what the hell is? I think you may have misread our posts. :-) The 470K is not an issue, as George and I both indicated. Regards, -- Al |
Perhaps the reason some of you are reporting sonic differences between using battery power and the wall wart is simply that volume levels haven't been precisely equalized for the comparison?
I realize that the LSA has an internal regulator, but no regulator is perfect, and so perhaps voltage differences between the outputs of the battery and the wall wart result in slight (fraction of a db) changes in attenuation, that need to be compensated for with the volume control.
To totally rule out the possibility that the sonic differences are attributable to volume changes, I believe that the levels should ideally be equalized to within around 0.1 db, which is probably impossible to do without instruments.
Best regards,
-- Al |
Anthony, no, I was assuming in my last post that both supplies are nominally 12VDC. However, neither is likely to be exactly 12VDC, and the two are not likely to be exactly the same. Also, the output voltage of the battery will change slightly as it discharges.
While the significance of those differences will be greatly reduced by the internal voltage regulator, as I said no regulator is perfect. So it seems conceivable to me that if the power source is changed but the volume control setting is not changed, there could be a tiny change in volume that would be subjectively perceived as a sonic difference.
Best regards,
-- Al |
Yes, the transport has no relation to the issue. The problem is that DAC's (and also CD players) usually have maximum output levels of 2 volts or more, which is unnecessarily high. Stereophile measured the maximum output level of your particular DAC as being 2.17 volts. As a rough approximation the outputs of tuners and other analog sources are often just 1/3 or so of that amount, which is about 10 db less.
Regards,
-- Al |
04-12-12: Almarg
I am not familiar with the characteristics of the 20 db version, but I would guess that it would present a load impedance that is significantly higher (and therefore even better). I should add to this statement, though, that since on occasion you have used your volume control as high as the 2 o'clock position, I'm uncertain as to whether or not the 20 db attenuator would still allow you to reach those same maximum volume levels, even at the maximum setting of the volume control. Regards, -- Al |
This is terrible, shocking, and saddening news. RIP, Paul. Our discussions have been a pleasure and a privilege.
-- Al |
Bill (Grannyring), a strong +1 to Swampwalker's comments. Very glad the surgery, and less importantly the system, have had fine results.
Best,
-- Al
|
So, yes there is a slight loss in max power, but do you think it would be relevant? My guess is that it would not be a problem, especially considering this statement by Jult52 in the other thread: My power amp has a continuous rated output of over 500w which I never fully max out being a law-abiding middle-aged person :) I like the power for the instantaneous peaks, especially at the onset of sounds and for the feeling of headroom. BTW, my rough estimate of 3db as the amount of the amp's power capability that would become unusable was based on the following: 1)1.3db due to the disparity between dac output voltage and amp input sensitivity. 2)0.5 to 1.0db due to tolerances or possible inaccuracies in those specs. 3)0.5 to 1.0db estimated margin in the recording, between the highest musical peaks and digital full scale. That's what I've actually seen on a number of commercial cd's, that I've analyzed with an audio editing program called Sound Forge. Best regards, -- Al |
Paul -- Yes, that is exactly correct IMO.
Although some might raise minor quibbles about the use of the word "ideal," because there will always be a limit to how severe a load a given buffer can handle without its performance degrading.
Best regards,
-- Al |
The (diode thingy) "diode effect" is bought around by very fast music transients from the source, CDP or phono, which can be in the order of 100's of volts per micro seconds (volts per uS) these happen as the name implies in micro seconds. While I have no reason to doubt the audibility of diode effects that may occur with many volume controls, I would respectfully point out that this explanation strikes me as fundamentally flawed. For line-level amplitudes, swings occurring at rates of 100's of volts per microsecond would represent spectral components (frequencies) at several tens of MHz. No source material and no source component will provide signal frequencies remotely approaching those rates. Not to mention that the amplifier, the speakers, and our ears would not respond to them even if they were somehow present. Regards, -- Al |
09-28-11: Banquo363
Am I right in believing that non-ideal impedance match between phono stage and LSA results in a attenuation of the high end? Perhaps it hasn't bothered me because of my old ears? Not necessarily. It depends on how the two impedances vary as a function of frequency. If neither varies significantly over the frequency range, there would be no adverse effects, just a slight gain reduction. I couldn't find any indications of the output impedance vs. frequency characteristics of your Allnic H-1200, but more often than not the most significant output impedance variation of a tube-type line-level component will be an impedance rise at deep bass frequencies, due to the output coupling capacitor that is commonly used. That would cause deep bass frequencies to be attenuated, assuming the input impedance of the destination component is reasonably flat as a function of frequency. What will attenuate high frequencies is a combination of high output impedance (at high frequencies) plus high cable capacitance (or, in theory, high input capacitance of the destination component, but that generally doesn't occur in practice, and I assume does not occur with the LSA). Regards, -- Al |
George, thanks for providing that background.
Paul, I've always interpreted the phrase "controlling the interconnect cables" to mean simply minimizing or eliminating their effects.
As you know, Ralph has made the point that in the case of a balanced cable driven by a balanced output circuit having very low output impedance, that is capable of driving 600 ohms without issues, and with a 600 ohm termination applied at the destination end, cable effects and cable differences will be completely eliminated.
I see no reason to doubt that, and I too have great respect for Ralph's opinions. But obviously most available equipment is not designed to those standards, which is one reason among a great many why assembling a system inevitably involves compromises and tradeoffs. The bottom line on those compromises and tradeoffs, though, seems to simply be the common sense notion that there are many paths that can lead to excellent results, including both passives and actives.
All technically explainable cable effects that I can envision are either directly proportional to or highly correlated with length, so I would not consider controlling the effects of a 1 meter or shorter cable to be a major factor in those tradeoffs.
IMO.
Best regards,
-- Al |
Given George's good answer about noise level, and his answer implying that the "line regulation" performance of the internal 5V regulator is good enough to reduce expectable variations in the 12V input to the point that they would not result in a perceptible volume change (which is certainly do-able in a quality design), I'm out of ideas as to what could account for sonic differences between battery and wall wart power (assuming that the wall wart being used is a linear supply, not a switching supply).
Best regards,
-- Al |
04-12-12: Gooddomino
My Krell amp has an input sensitivity of 1.34 V RMS and gain of 26db. Looks like I need an amp with lower input sensitivity . Not sure how the gain number factors in. Sensitivity (the input voltage that will result in the amp putting out its maximum output power) is not directly relevant. What is relevant is the gain of the amp (the relation between output voltage and input voltage). Comparing sensitivity numbers is only meaningful if put in the context of the power capability of the amp. For example if two amps both have an input sensitivity of 1 volt, but one of them is rated at 50 watts and the other at 500 watts, the 500 watt amp will produce much more volume in response to a given input signal than the 50 watt amp will produce. The suggestion George made earlier about trying a pair of Rothwell attenuators seems to me to be well worth trying. They would be connected directly to the input jacks of the amp, with no intervening cable. In conjunction with the 47K input impedance of your amp, the 10 db version would result in the LSA seeing a load impedance of about 30K, which is on the low side but I suspect would still be ok, given that it won't vary significantly with frequency and given the 620 ohm output impedance of your source. I am not familiar with the characteristics of the 20 db version, but I would guess that it would present a load impedance that is significantly higher (and therefore even better). Regards, -- Al |
