Announcement

Collapse
No announcement yet.

Is desire for more powerful amps overrated?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #61
    Originally posted by atmasphere View Post

    I am sure in your second sentence you meant voltage and not power.
    No, I meant power. And explained what I was getting at.
    Originally posted by atmasphere View Post

    But since that did come up, turns out there is more than one pathway to neutrality:

    http://www.atma-sphere.com/Resources...ier_Design.php
    Interesting thesis. There are many parts to agree with, and some to disagree with. Some of what he says I've just said also in a slightly different way.
    Originally posted by atmasphere View Post

    There are two issues at play here. The first and probably more important is the simple fact that our ears convert all forms of distortion into a tonality- the richness of tubes caused by the 2nd or 3rd harmonic, the harshness of solid state by the higher orders and so on. The ear/brain system has tipping points where sounds are often processed differently; in the case of distortion, the ear can favor tonality caused by distortion over actual FR. Since no loudspeaker made also has flat frequency response, a system that is less than flat can sound far more neutral if the ear isn't responding to distortions to which it is keenly sensitive (the higher ordered harmonics are used by the ear to sense sound pressure so the ear is more sensitive to them than lower orders).
    IMO, the above is a fairly sizable over-simplification of the distortion audibility problem.
    Originally posted by atmasphere View Post
    Now the masking principle plays a role here- the presence of the 2nd and 3rd harmonics mask the presence of higher orders- that's why I said that 'the ear isn't responding' in the paragraph above. Its this masking principle which is why tubes sound smoother (and also at clipping they are making less higher ordered distortions).
    Only works when considering masking with single pure tone stimulus. Unfortunately, the principle is complicated by harmonic masking by broader band stimulus like music.
    Originally posted by atmasphere View Post
    Add to that the simple fact that if you make any amplifier work harder for its living (i.e. drive a 'difficult' load) its going to make more distortion, which will be audible as harshness and brightness (and loss of detail). This is easy to see in the specs of any amplifier.
    The above is not true, and just as a sanity check, the very first amplifier performance tests I googled (not having actually read the test ever before) shows the opposite is true. See below, which shows as power goes up distortion goes down, which is actually very common, especially with SS amps employing feedback.

    ​​
    Originally posted by atmasphere View Post
    People like to say the difference is 'negligible' but the distortions involved are those pesky higher orders, so they aren't 'negligible'. We really should have a weighting system of harmonic distortions based on the ear's sensitivity to them...
    Some people might say that typical distortion in amps is negligible, but again, that would be an over-simplification of the problem of distortion audibility. Nearly all amp distortion spectra is comprised of both low and high order harmonics, not just high-order. In fact, the typical spectrum shows low-order harmonics as significantly higher, as in this case which shows 3rd being dominant, with the next highest being 5th at more than 10dB below 3rd:

    Originally posted by atmasphere View Post
    Now the 2nd bit: voice coils can heat up on transients enough to produce easily measured reduced dynamic range (some studies have produced 7dB results...). To some extent, this can be reduced by voice coils of larger diameter, but on lower efficiency speakers (ESLs excluded) this is simply a thing. When the voice coil heats up, the impedance also goes up and a voltage source amplifier can't make as much power into the speaker. That this can happen fast enough on musical transients should not be a surprise- light bulbs were used by Fairchild in their famous compressors in much the same way.
    Again, true, but over-simplified. First, voice coil heating related compression is real, but it's audibility is dependent on many things. The heating effect is dependent on the RMS value of the applied signal, not the peak value. RMS is, of course, also called "heating value". That's true with light bulbs, and that is actually why Fairchild and others (UREI) used them as a method of RMS detection and gain control in some (not all) of their products. The RMS value of a signal also is the dominant factor in loudness determination by human hearing. RMS detection also includes a time constant, how long a heating signal is applied before a stable amount of heating is reached, which depends on the thermal characteristics of the individual driver. It's not just a matter of how much additional resistance is added by heating, but how long it takes to get to that level, and how long it takes to recover from it by cooling. As in dynamics processor design, the attack and release time constants haver more to do with audibility that actually attenuation.
    Originally posted by atmasphere View Post
    There is no way that more power solves this problem- it simply makes it worse.
    Yes, I did say that.
    Originally posted by atmasphere View Post
    Lower efficiency speakers are a result of inexpensive power made available by solid state amps. The race for higher power has led us nowhere- when cheap power became available, right away the speaker industry started making less efficient speakers, the reason being that low efficiency speakers are a lot less expensive to produce.
    I agree there is a relationship between available amplifier power and speaker design, but I think of it as the other way around. Lower efficiency speakers came first, because the designs easily solved a number of design problems, but the there was a need for higher power to make those designs practical.
    Originally posted by atmasphere View Post
    The woofers in my speakers are 97.5dB and go to 20Hz with ease in my Classic Audio T-3s, but they also cost $2000.00 each. If 10dB less output, they might well be only $200.00. Yet somehow speaker manufacturers have worked it out to be able to charge the same price for the finished product- essentially the industry is just more profitable. Nothing wrong with that, but there's also nothing wrong with knowing it either!
    We should be cautious about coupling $$ to dB. The rule gets broken routinely. The problem with sub design isn't just about how low they go (and stating a single figure response is meaningless, BTW. 20Hz at what relative level, maximum SPL and distortion?) it's about how low, how clean and at what maximum SPL...and then there's the real limiter...the room.
    Originally posted by atmasphere View Post
    IMO high end audio is all about getting electronics to sound real. In that regard I simply don't see the need for higher power, lower impedances and harder to drive speakers. Just doesn't make sense, not if getting it to sound real your goal.
    Would it make sense if the speakers you want to use are a less than efficient design? Or if the distance to your LP is large?

    IMO, audio in general is about picking your compromise. High end audio is about reducing the constraint of one compromise: cost, and hopefully optimizing others as a result. That doesn't always happen, but that's fine. I'm more bothered by the tendency to create and perpetuate myth as fact in that portion of the market.

    Comment


    • #62
      Of course its all about speaker efficiency, room size and listening level. Its always been about that.

      The problem is that the "it sounds fine" observations are pretty lame justifications for setting power levels. Our ears can mask problems unless we have been trained to recognize power limitation issues.

      It all starts with your software. What kind of music source are you using and what are the dynamic requirements for reproducing that content?

      A LP with a small Jazz group playing in a club is not exactly one of your better examples of dynamic content. It can be but you need something with extreme low frequency content. The lowest frequency demands recorded well will require the most amount of power to reproduce.

      A recording of a pipe organ or a full orchestra that has been recorded well and mastered with full range demands in tact are some examples of software that will test your system, and require the most amount of power. The instantaneous peak power demands on a great recording can require much more power than many of us realize.

      If you want to reproduce a great recording at a realistic listening level you need a lot of power. How many of us live with limited low frequency content? I believe the answer to that question is most of us live with restricted low frequency content.

      Reproducing a kick of a drum, recorded well, requires that we move a lot of air. High quality systems do a decent rendition of extreme low frequency content but it is muted by design. The speaker manufacturer knows that it takes extreme amounts of power to reproduce low frequency content.

      It seems to me that few speakers are designed to reproduce that kind of sound. Only the extreme examples of speaker design are designed to reproduce extreme low frequency content. The Infinity IRS V comes to mind as a speaker designed with full low frequency content. We are talking about a lot of speaker cone area driven by thousands of watts of power to energize a listening room with full low frequency content.

      That kind of speaker is what it really takes to reproduce extreme low frequency content. I am not knocking other speaker designs, it is just a decision made when designing a speaker. There are also speakers that couple to our listening spaces to try and reproduce the LF content. Corner loading of a horn design, like with the Klipschorn, is another approach to dealing with the LF demands.

      It takes a lot of power to deal with low frequency reproduction. If you want to feel the kick drum in your chest, it takes power. And lots of it.
      Life is analog...digital is just samples thereof.
      https://www.edsstuff.org

      Analog: VPI Prime, 3 Belt Drive, Dual Pivot, ADS, Ortofon Windfeld Ti, Liberty B2B-1, Sky 30 Trans, Stereo Squares Dust Cover
      Amp: Rogue RP-9 + Schiit Loki + Parasound A21
      Speakers: Magnepan MMC2 + DW-M + REL T5/i
      Digital: TASCAM UH-7000, Pioneer DV-47Ai, NAD C446, Schiit Modi Multibit
      Headphones: Stax Lambda Pro, Focal Clear, Schiit Lyr 2
      Wires: Kimber Hero + 8TC + AQ Leopard

      Comment


      • Tech7738
        Tech7738 commented
        Editing a comment
        I agree with everything, except "Life is analog". Analog: something that is similar or comparable to something else..analogous. Life (i.e. light and acoustic sound) isn't analog, it's the real thing. Analog signals are analogous to life. See how nit-picky I am? Just kidding, of course. Great post.

    • #63
      Originally posted by Tech7738 View Post
      The above is not true, and just as a sanity check, the very first amplifier performance tests I googled (not having actually read the test ever before) shows the opposite is true. See below, which shows as power goes up distortion goes down, which is actually very common, especially with SS amps employing feedback.
      I do simplify quite a lot in order to speak and write in layman's language- we're on the same page here. But in the quote above, I think you misinterpreted my comment. I know the phenomena you are addressing and I've seen it quite a lot as well. But I was not referring to that- simply that if you put any amplifier on a lower impedance load and/or obnoxious phase angles and the like, it will have measurably higher distortion. IOW look at the 4 ohm distortion spec of any amp as opposed to what it does into 8 or 16 ohms.

      IOW this is simply an argument for easier to drive loudspeakers, since distortion that arises from amps driving lower impedances is audible.

      Comment


      • #64
        Originally posted by atmasphere View Post

        I do simplify quite a lot in order to speak and write in layman's language- we're on the same page here. But in the quote above, I think you misinterpreted my comment. I know the phenomena you are addressing and I've seen it quite a lot as well. But I was not referring to that- simply that if you put any amplifier on a lower impedance load and/or obnoxious phase angles and the like, it will have measurably higher distortion. IOW look at the 4 ohm distortion spec of any amp as opposed to what it does into 8 or 16 ohms.
        Another sanity check was done, here's a few I found (I looked at Stereophile, my only criteria was amp with full sets of THD vs power graphs. The site is a bit inconsistent that way:
        https://www.stereophile.com/content/...r-measurements
        4 ohm THD is lower than 8 ohm

        https://www.stereophile.com/content/...r-measurements
        8 ohm THD is lower than 4, and 4 is lower than 2

        https://www.stereophile.com/content/...r-measurements
        The 8 ohm curve and 4 ohm curve cross at certain points, but the bulk of the 4 ohm curve is lower.

        Perhaps it's not so simple that it can be generalized? Perhaps "if you put any amplifier on a lower impedance load...it will have measurably higher distortion" is to general and restrictive?

        Comment


        • #65
          In my room I have tried a few amps and each has a level of attack they can show at my place. The mono amps I use now are extreme but never show me signs of clipping
          Or sonic restraint . Even smaller speakers at friends places have shown improvements from small amps to a larger one. Yes some are sonic traits of the amps but a more powerful amp at times yields a more power stage
          again there are many variables in this both math and likes are involved.
          Ps my woofer towers alone each have 2000 watts
          but you don't see them if setup well and they add to the stage as needed meaning they blend into the other towers.
          speakers cust infinity IRS V , new caps and LPS , magnets etc.
          mark levivson pre no 26 amps no 33
          digital three cust servers , win ser 2016 , AO
          Dacs lampi various

          Comment


          • #66
            Originally posted by Tech7738 View Post
            Another sanity check was done, here's a few I found (I looked at Stereophile, my only criteria was amp with full sets of THD vs power graphs. The site is a bit inconsistent that way:
            https://www.stereophile.com/content/...r-measurements
            4 ohm THD is lower than 8 ohm
            Huh? From that article:
            The First Watt SIT-3 offered levels of distortion that remained relatively constant with frequency but that increased into lower impedances (fig.6).
            (emphasis added)
            Did I miss something?


            https://www.stereophile.com/content/...r-measurements
            8 ohm THD is lower than 4, and 4 is lower than 2

            https://www.stereophile.com/content/...r-measurements
            The 8 ohm curve and 4 ohm curve cross at certain points, but the bulk of the 4 ohm curve is lower.

            Perhaps it's not so simple that it can be generalized? Perhaps "if you put any amplifier on a lower impedance load...it will have measurably higher distortion" is to general and restrictive?
            Generally speaking, the final link you provided shows lower distortion if the load impedance is higher. Am I missing something? These links seem to support my earlier comment...

            Comment


            • #67
              Originally posted by atmasphere View Post

              Huh? From that article:

              (emphasis added)
              Did I miss something?
              Apparently. I superimposed the two THD graphs for you to make it more clear.
              Click image for larger version

Name:	firstwatt.jpg
Views:	84
Size:	54.1 KB
ID:	116942

              Originally posted by atmasphere View Post
              Generally speaking, the final link you provided shows lower distortion if the load impedance is higher. Am I missing something? These links seem to support my earlier comment...
              Hardly.

              Here's a couple more superimposed:

              Click image for larger version

Name:	verity.jpg
Views:	73
Size:	61.3 KB
ID:	116943
              Note the point at which 8 ohms is higher THD than 4 ohms, though for most of the curve 8 ohms is lower.


              And the Pass Labs:
              Click image for larger version

Name:	Pass-Labs-XA200-8.jpg
Views:	72
Size:	66.5 KB
ID:	116944

              My point in the above is that there is not the consistency one would hope for in the statement: " "if you put any amplifier on a lower impedance load...it will have measurably higher distortion". As you can see, it's not true at all for the First Watt, and only true for part of the power curve for the other two. That would make the statement flawed, and a bit too general. The lower impedance (theoretically the "more difficult" load) does not show consistently higher distortion.

              edit: I think I fixed the broken image link issue. If not, somebody let me know.
              Last edited by Tech7738; 06-04-2019, 09:33 PM.

              Comment


              • #68
                Originally posted by Tech7738 View Post
                Apparently. I superimposed the two THD graphs for you to make it more clear.



                Hardly.

                Here's a couple more superimposed:

                Note the point at which 8 ohms is higher THD than 4 ohms, though for most of the curve 8 ohms is lower.


                And the Pass Labs:


                My point in the above is that there is not the consistency one would hope for in the statement: " "if you put any amplifier on a lower impedance load...it will have measurably higher distortion". As you can see, it's not true at all for the First Watt, and only true for part of the power curve for the other two. That would make the statement flawed, and a bit too general. The lower impedance (theoretically the "more difficult" load) does not show consistently higher distortion.
                The links/image sources appear to be broken.

                Why does JA conclude that the amp has more distortion into lower impedance in the first example you posted? He seems to contradict you in this regard.

                Comment


                • #69
                  Originally posted by atmasphere View Post

                  The links/image sources appear to be broken.
                  They work fine for me. Links are to the audionirvana server.
                  Originally posted by atmasphere View Post
                  Why does JA conclude that the amp has more distortion into lower impedance in the first example you posted? He seems to contradict you in this regard.
                  No idea. The graphs don't agree with that at all. All I'm doing is going by the published Audio Precision graphs.

                  Here are direct links to the superimposed graphs, in case you still can't see them.:

                  https://i.ibb.co/Z6Nzg7g/firstwatt.jpg

                  https://i.ibb.co/Zhr3N0c/verity.jpg

                  https://i.ibb.co/BZWzXgV/Pass-Labs-XA200-8.jpg

                  Comment


                  • #70
                    Originally posted by Tech7738 View Post
                    They work fine for me. Links are to the audionirvana server.


                    No idea. The graphs don't agree with that at all. All I'm doing is going by the published Audio Precision graphs.

                    Here are direct links to the superimposed graphs, in case you still can't see them.:

                    https://i.ibb.co/Z6Nzg7g/firstwatt.jpg

                    https://i.ibb.co/Zhr3N0c/verity.jpg

                    https://i.ibb.co/BZWzXgV/Pass-Labs-XA200-8.jpg
                    Now for some reason the images that were broken yesterday are rendering properly...

                    I found out why JA made that comment. Scroll down to Figure 6:
                    https://www.stereophile.com/content/...r-measurements

                    We have to keep in mind here that 2.83V into 4 ohms is 2 watts, not one. When you start to unpack this stuff, there are some holes (IMO) of how JA has done his measurements. Wouldn't it be a better idea to see what the 1 watt power is since that first watt is pretty important, especially if you call your product First Watt?? I'd also like to see the harmonic spectra to at least 7KHz from a 50Hz tone, on account of Fletcher Munson. This is the area where 'brightness' and harshness can occur- the most sensitive region of human hearing, and no data from his tests...

                    Comment


                    • #71
                      Originally posted by atmasphere View Post
                      Wouldn't it be a better idea to see what the 1 watt power is since that first watt is pretty important, especially if you call your product First Watt?? I'd also like to see the harmonic spectra to at least 7KHz from a 50Hz tone, on account of Fletcher Munson. This is the area where 'brightness' and harshness can occur- the most sensitive region of human hearing, and no data from his tests...
                      So now we'd like to be more specific? No argument there, in fact, it was the non-specificity of the original generalization that I was objecting to in the first place. If you really want to see what's going on with THD, it's a 3d graph of power vs frequency vs level. And that won't quite show you everything either since you need the spectrum as well. And that won't show you audibility either because masking can't be accounted for.

                      I wouldn't focus on harmonic content of 50Hz in the 7kHz region though. It will be extremely low in just about all cases, and masked by music content with fundamentals and harmonics in that region that dominate unless the the distortion is extremely high, like that created by hard clipping.
                      For example, most would consider pushing 3dB past a hard clipping point as extreme distortion, right? Third harmonic is less than 20dB below the fundamental, clearly audible, etc. Below is the spectrum of a 50Hz sine wave pushed 3dB above hard clipping:
                      Click image for larger version

Name:	50hz 3db clipped.jpeg
Views:	65
Size:	95.2 KB
ID:	117044
                      Note that the 3rd harmonic is -17.5dB below the fundamental, and would be clearly audible, but the harmonics produced at 7kHz and above are all -86.5dB below the fundamental, which is approaching the noise floor of 16 bits. The area that would produce obvious audibility is all below 1kHz. And this is produced by absolute hard clipping. (Notes: the odd shape between harmonics and below 50Hz is a result of FFT windowing. The red hash marks were added for ease of comparison. The 0dB reference is 0dBFS.)

                      I would look at high frequency intermodulation distortion that produces non-harmonically related products between 3kHz and 10kHz.

                      We're now going to the polar opposite of "if you put any amplifier on a lower impedance load...it will have measurably higher distortion", (which was what I was pointing out) and moving toward specific characterization. As testing gets more specific and detailed, amplifier differences begin to pop out.

                      Relating audibility to single-figure THD, even if graphed over a range of levels, is folly. But "real" distortion profiling is also indigestible by most who read reviews and specs, and quite time consuming to perform.

                      Comment


                      • #72
                        Originally posted by Tech7738 View Post

                        I would look at high frequency intermodulation distortion that produces non-harmonically related products between 3kHz and 10kHz.

                        Seems a good idea, even slightly less IMD is easily heard. I don't recall seeing IMD vs load impedance though.

                        Comment


                        • #73
                          Originally posted by atmasphere View Post
                          Seems a good idea, even slightly less IMD is easily heard. I don't recall seeing IMD vs load impedance though.
                          Note that I specifically mentioned high frequency IMD with multiple tones. SMPTE IMD is not particularly useful in SS circuits, more pertinent in tube circuits. SMPTE IMD usually tracks THD.

                          You won't see IMD vs load unless someone decides to do it. Completely possible with anything from free all the way to AP analysis systems. However, multi-tone IMD at high frequencies isn't standard. The best you could do is two tones a couple kHz apart without a more involved procedure. Even that would be an improvement over nothing.

                          Comment


                          • #74
                            Me I would not own a bench mark hahaha. When I went to the Nyc audio show one room was by technics I think. They had all there stuff and was called digital
                            me being ignorant ask is this class D ? He said no it's digital lol. It was not music playing it hurt the listen too.
                            Not did not have peaks or lacked jump or bad tone
                            or even did not make me feel it's timbre was nuts.
                            But it was not very good at producing real sounding music.
                            So tech can we measure bad sound lol.
                            I agree we can measure all we hear. But I don't feel we know enough to equate all it is.
                            An amp of extremely low distortion some feel is a dry sounding amp. Me I don't know why but I do of it's music to me.
                            speakers cust infinity IRS V , new caps and LPS , magnets etc.
                            mark levivson pre no 26 amps no 33
                            digital three cust servers , win ser 2016 , AO
                            Dacs lampi various

                            Comment


                            • #75
                              once again, when it comes to measurements:

                              Click image for larger version

Name:	unnamed.png
Views:	56
Size:	17.2 KB
ID:	117104
                              Steve Lefkowicz
                              Senior Associate Editor at Positive Feedback
                              --------------------------------------------------------
                              http://www.audionirvana.org/forum/ti...ounding-system

                              Comment

                              Working...
                              X