Bombaywalla, thanks for the additional thoughts. However, while I agree with your math, I don't completely agree with the conclusions you reached. As I'm sure you recognize, my hypothetical example involving 200 watts at 200 kHz, as well as the figures you cited in connection with Bruce's Ref150 of 150 watts at 66 kHz, are not remotely within the realm of real world operating conditions for a home audio system. Or at least a home audio system that is not in the process of self-destructing due to some unimaginable defect-induced oscillation :-) A key point to keep in mind is that under realistic operating conditions bandwidth and risetime should not be and in general will not be limited by slew rate, as I indicated in my long post dated 2-6-15 (see more on that below). And as I indicated in my previous response to Bruce: ... a basic point in my analysis was to show that even relative to VERY conservatively drawn thresholds the speed-related parameters of the Spectral amps are **in themselves** extreme overkill. While at the same time citing the possibility that the overkill might provide indirect benefits in the design. Consider "VERY conservatively drawn thresholds" to mean parameters that are so extreme as to be well beyond the bounds of real world operating conditions. As you may realize but others may not, slew rate should be fast enough to avoid any possibility of the amplifier's output slewing under any realistic operating conditions. While bandwidth (and risetime) limitations mean that frequency components above a certain frequency are attenuated or filtered out, slewing results in essentially all frequencies that represent sonically meaningful information being filtered out and/or distorted. In a sense it can therefore be considered analogous to what happens in hard clipping, although of course that relates to amplitude limitations rather than speed limitations. I note, btw, that the ARC Ref 150 has a specified 3 db bandwidth of 120 kHz (at 1 watt, with an 80 kHz "power bandwidth" also being specified), and JA's measurements indicate a bandwidth of approximately 100 kHz (for an 8 ohm load connected to the 8 ohm tap), with excellent reproduction of a 10 kHz square wave (under small signal conditions, as noted under the graph). So while I don't think we can completely rule out the possibility of audible phase shifts occurring in the upper treble region under some circumstances, given the bandwidth of the Ref150, I would not expect that to be as much of an issue as your conclusion would seem to indicate. And to put that into context, I'll point out that more than a few other very highly regarded amps, including some tube amps costing vastly more than the ca. $13K Ref150, have bandwidths which are considerably more limited. Best regards, -- Al |
Ptss 02-05-15
SPECTRAL claims faster today.... Does it make any difference to the music? Ptss, if I recall correctly you have a DMA-180. I compared the specs on that amp with those of the currently produced DMA-200S S2, DMA-260 S2, and DMA-300. All of the speed-related specs (frequency response/bandwidth, risetime, settling time, and slew rate) on all four amps are identical, except that the DMA-300 has a slew rate of 800 V/us while the others are 600 V/us. Most of the other specs are also very similar, the most notable differences being that the three newer amps have a much lower input impedance (10K) than the input impedance of your amp (100K), which could be problematical with some preamps, and the three newer amps have rated maximum power capabilities which increase into decreasing load impedances more slowly than in the case of your amp. The bottom line, IMO. Regards, -- Al |
Hi Peter (Csontos), Yes, slew rate, risetime, and bandwidth will in the case of many designs extend well beyond the point of being overkill with respect to effects that may have DIRECT audible significance. The Spectral amplifiers being extreme examples in that respect. As I'll get into in a moment, though, that does not necessarily mean that the only benefit is to those who write marketing literature. But first for some quantitative perspective on these numbers: A very conservative (i.e., safe) rule of thumb for bandwidth, that would rule out the possibility of perceptible phase shifts within the audible spectrum, and that would take into account the possibility that, as some have claimed, frequencies well above 20 kHz may be sensed by some listeners when accompanied by lower frequencies, would be to multiply 20 kHz by a factor of 10, meaning 200 kHz. Then consider that the relation between risetime and 3db bandwidth (i.e., the frequency at which response has rolled off by 3 db) is, for a first order (6 db/octave) rolloff: risetime = 0.35/bandwidth where the units are chosen consistently between the two parameters (i.e., seconds and Hz, or milliseconds and kHz, or microseconds and mHz). Based on that relation, a first order response that is down 3 db at 200 kHz corresponds to a risetime of 1.75 us (microseconds), very close to the 2 us figure you cited. Slew rate, which is the fastest rate at which the amplifier output can change (presumably under large signal conditions), should be fast enough to provide a wide margin relative to values that might limit the other two parameters. If we consider the somewhat extreme example of a 200W amp, 200 watts into 8 ohms corresponds to 40 volts rms, which for a sine wave corresponds to 113 volts peak-to-peak. The half-period corresponding to a frequency of 200 kHz (i.e., the time required for a peak-to-peak swing at that frequency) is 2.5 us. 113 volts/2.5 us is about 45 volts/us. A slew rate of 100 volts/us, which Ralph (Atmasphere) rightly characterized as "fast," seems comfortable relative to that figure, considering especially that 200 watts at 20 kHz, much less 200 kHz, is an unlikely scenario :-) Now consider in those contexts the specs on the four Spectral amps I referred to: Bandwidth (3 db): 1.8 mHz Risetime: 0.4 us Slew rate: 600 v/us (800 v/us in one case) Obviously way overkill relative to any conceivable effects that are DIRECTLY audible. So why have I put the word directly in caps? Well, I would presume the designers feel that by designing ultra-fast circuitry they can avoid or minimize effects which may be audibly significant. For example, they may be able to realize the benefits of increased amounts of negative feedback while avoiding or minimizing what would normally be its adverse effects, such as transient intermodulation distortion. Or the higher speed circuitry might help to minimize crossover distortion, or the effects of unwanted energy storage in devices, etc. So what can be said about that? Well, its an approach, and a philosophy. As is usual in audio, how good or bad the results are will depend on quality of implementation, system matching, and listener preference. My understanding is that for the most part sonically, slew rate correlates to bass performance and rise time to treble. This is what I've been told by the tech who's worked on some of my amps. That can probably be said to have a vaguely definable basis in situations where risetime and slew rate are slow enough to verge on having DIRECT audible significance, given as I said that risetime is inversely related to bandwidth (and thus potentially to upper treble extension), and bass frequencies are generally present at much greater amplitudes than upper treble frequencies (slew rate probably being best considered as a large signal parameter, as I said earlier). But the statement strikes me as a crude oversimplification at best, and one which does not have much relevance in situations where both parameters have comfortable margins relative to audible thresholds. Best regards, --Al |
Bruce (Bifwynne), a basic point in my analysis was to show that even relative to VERY conservatively drawn thresholds the speed-related parameters of the Spectral amps are **in themselves** extreme overkill. While at the same time citing the possibility that the overkill might provide indirect benefits in the design.
The numbers you cited for your ARC amp don't seem to me to be unreasonable in any way, they just represent a different philosophy and a different approach, which result in less margin (but arguably still very reasonable margin) in those particular parameters.
Mapman, re your question, I'm not really sure. I too believe that there is a strong relation between resolution and holography/dimensionality etc., but my instinct is that linearity, distortion, and noise performance more often than not tend to affect resolution more significantly than the speed-related parameters. Just MHO.
Best regards,
-- Al
|
Al; where the heck did you get this knowledge? I'm an EE with extensive analog and digital design experience (for defense electronics, not for audio), and I've been an audiophile for about 35 years. Audio has always been of interest to me from a technical standpoint as well as a musical standpoint. Best regards, --Al |
JJ's doesn't have any locations in my area. But before playing a recording that might call for a sub (which doesn't happen too frequently with my speakers), I can always drive over to a Subway location if need be. :-)
Best regards,
-- Al
|
Bruce, I agree with everything in your post, and also with Bombaywalla's last post. The one about slew rate, that is; I have no experience with Quiznos :-) Upon careful reading I don't see the two posts as being inconsistent with each other.
Best regards,
-- Al
|
Thanks, Ptss. I appreciate the nice words and the suggestion. And I've noted in the past your frequent advocacy of the importance of addressing power quality and power-related issues.
I'll keep your specific suggestion in mind, for a future time. Next on my list, though, as you've seen in the DEQX thread you initiated, is purchase of an HDP-5 in the next month or two :-)
Best regards,
-- Al
|
02-18-15: Csontos
Al, as you and Ralph are obviously the two most technically knowledgeable, I think it most appropriate for the two of you to also define/describe amplifier 'speed' from a subjective/sonic perspective. From a subjective perspective, I think of "fast" as referring to the ability to reproduce musical transients in a "clean" and accurate manner, as opposed to a "sloppy" and inaccurate manner. As ZD said on 2-9-15, "you need to listen to how the system is reproducing the attack, resonance and decay of whatever instruments are on the recording." I would emphasize, though, that speed from a subjective standpoint and speed as characterized by risetime, bandwidth, and slew rate are two different things, which do not have a necessary or direct relation. It seems entirely conceivable to me that an amp having a 100 kHz bandwidth could be subjectively "faster" than one having a 1 mHz bandwidth. Although in many cases there may be an indirect relation between faster bandwidth/risetime/slew rate and subjective speed. For example, use of a given amount of feedback in a really high speed circuit may not result in as much Transient Intermodulation Distortion compared to its use in a slower circuit (as Ralph, Psag, and I alluded to earlier). Less TIM distortion meaning more accurate reproduction of musical transients, by definition. BTW, re "the two most technically knowledgeable," let's not overlook Bombaywalla, who is certainly no slouch when it comes to EE technical matters, and perhaps some of the others who have responded as well. Best regards, -- Al |
02-19-15: Csontos
If the relationship between technical parameters and subjective perception is indirect or vague, then how does someone set out to design a 'subjectively fast' amp? Is it just a crap shoot? Technically, what descriptors must an amp possess in order to facilitate subjective speed? Isn't this the goal? Thanks for the nice words in your post, Peter. I'm not sure how to answer your question, though. Consistent with Ralph's comment, there are simply too many variables that can be expected to be involved to be able to formulate a meaningful general purpose answer. It stands to reason that wide bandwidth and fast risetime and slew rate will tend to be helpful, but only indirectly if at all once they get above a certain point (for example, the 200 kHz bandwidth figure I cited earlier, corresponding to 10x the nominal upper limit of audibility). As with most sonic characteristics of audio components, I would expect accurate reproduction of those sonic attributes which contribute to a realistic perception of "speed" to be more than anything else a function of the knowledge, experience, and thoroughness of the particular designer, and of his or her priorities in addressing the innumerable tradeoffs that enter into any design. Best regards, -- Al |
Great post just above, ZD. I agree completely.
Regarding your post just above that one, I think that Ptss's comment you quoted was addressing my post, not yours. I don't have first-hand knowledge of recording industry practices, or of the design of the elaborate mixing consoles that are commonly used. However my understanding is that the widely held notion that a recording stands a 50-50 chance of having correct polarity, and that inverting the signal will correct the polarity of those recordings which do not have correct polarity, is a misconception. Especially in the case of recordings having many performers, and given that the majority of recordings (both classical and popular) tend to be heavily multi-mic'd.
My understanding is that the pickups from each of those many mics will typically be subjected to processing in the consoles which differs from mic to mic, making it likely that in the final mix some of the instruments and/or singers on the recording will have one polarity, and other instruments and/or singers will have the opposite polarity.
Unless, that is, the recording was produced using "purist" techniques, i.e., a minimal number of microphones and minimal post-processing. That being the exception rather than the rule, in most genres including classical.
Best regards,
-- Al
|
Regarding absolute phase/polarity, I would make two points:
1)It should be kept in mind that if a preamp provides a polarity switch, when the setting of that switch is changed not only is the polarity of the signal being inverted, but the circuit configuration that is being used within the preamp is being changed. In the case of a preamp having a balanced internal signal path, such as Ralph's designs, I would expect that to not be of much if any significance from a sonic standpoint, since the polarity change can be implemented in balanced designs by simply interchanging signal connections somewhere in the signal path. However in the case of preamps having an unbalanced internal signal path, I wouldn't rule out the possibility that the preamp's sonics could change noticeably when the setting of that switch is changed. Perhaps more noticeably than the resulting "correction" of the polarity of the recording.
2)I disagree with the notion of a 50/50 split, because I believe that for most recordings neither setting can be thought of as being correct. My suspicion is that on 90+ percent of recordings the whole concept of maintaining absolute phase or polarity is meaningless. The great majority of pop recordings, and the majority of classical recordings as well, are captured with a multitude of microphones (dozens in the case of many orchestral recordings), and then the signals from all of those mics are processed through elaborate multi-track consoles where they are mixed, equalized, compressed, limited, and subject to who knows what additional effects. The end result, when it comes to absolute phase or polarity, is inevitably in those cases a random hodge-podge of different sounds with different phases.
So if one setting of the polarity switch on a preamp sounds better than the other on a particular recording it figures to be more likely the result of random synergy and subjective preference than the result of the system preserving the "polarity" of the recording.
I therefore agree with Ralph that the benefit of a well implemented polarity changing function in a preamp will primarily be to the unfortunately very small percentage of recordings that are engineered using minimal/purist mic techniques, and that are engineered with minimal post-processing. But then again, those tend to be the kind of sonically high quality recordings that many audiophiles (including me) particularly seek out.
Regards,
-- Al
|
Ptss, amplifier damping factor is usually specified based on the assumption that speaker impedance is 8 ohms. So:
DF = (8 ohms)/(output impedance, expressed in ohms)
and equivalently,
Output impedance (expressed in ohms) = (8 ohms)/DF
As you can see, DF itself is not measured in ohms or any other units, being a ratio of two quantities that are measured in ohms.
In the case of tube amps that have output transformers and provide both 4 ohm and 8 ohm taps, DF is sometimes specified separately for the 4 ohm tap. Although the 4 ohm and 8 ohm DF’s will usually not be greatly different, as the output impedance of the 4 ohm tap can be expected, roughly speaking, to usually be in the general vicinity of half of the output impedance of the 8 ohm tap.
Regarding the frequency upon which the specified DF is based, it probably doesn’t matter in most cases because in contrast to speaker impedances amplifier output impedance will not usually vary enough over the frequency range to matter.
Enjoy the brandy! Best regards,
-- Al
|
