1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Skeptico Saloon: An Objectivist Joint

Discussion in 'Sound Science' started by gnarlsagan, Jun 27, 2013.
100 101 102 103 104 105 106 107 108 109
111 112
  1. miceblue
    Yeah that's what I thought too. Usually the gain is fixed at, say 2.5x in the Objective 2. Adjusting the volume knob wouldn't adjust the 2.5x gain, but the gain switch will. So now I'm a bit confused when cjl says the volume knob adjust the gain.
    Vout/Vin = gain

    Volume knob in a low position:
    2.5 V / 1.0 V = 2.5

    Volume knob in a higher position:
    5 V / 2.0 V = 2.5

    5 V > 2.5 V = more power
    2.5 == 2.5 same gain

    Volume knob in a low position, low gain:
    2.5 V / 1.0 V = 2.5

    Volume knob in the same low position, high gain:
    5.0 V / 1.0 V = 5.0

    5.0 > 2.5 V = more power
    5.0 > 2.5 = more gain

    I almost never touch the gain button/switch on an amplifier; usually the lowest gain setting on a given setup is all that I need.

    I do see the point cjl is trying to make though. If I play a loud song and switch to a quiet song, the song itself is quieter, so the input signal to the amp, and thus the output, will be smaller accordingly.

    The question is then, how can I predict how much I would need to turn the volume knob to output the same power, and thus SPL from the used speakers/headphones, to match that of the louder track (without using ReplayGain)? Earlier I modeled a specific situation for the OPPO HA-2 driving a HiFiMAN HE-560 in a practical scenario. My preliminary calculations seemed to fit my subjective testing, but the model may have been incorrect from the start since I assumed that with a given volume knob position, the amp would output the same power regardless of the input signal (loud to quiet track in this case).
  2. Steve Eddy

    If you just look at an amplifier as a "black box" with an input and output, you can look at it either way. I was just speaking to how it's actually implemented in a real amplifier, i.e. the gain is fixed and the volume control just attenuated the incoming signal.

    With the volume control in a given position, the amp will still only output power depending on the level of the input signal. If the input signal is lower, output power will be lower.

    Without something like ReplayGain, no way to predict it unless you just adjust volume by ear and note the posupition of the volume knob for each song.

  3. miceblue

    Still without ReplayGain, is there a way to predict how much power an amp would output provided that you have the songs' RMS values, the amp's gain (gain as in an op-amp gain, R2/R1 + 1), as well as the load the amp is driving (i.e. headphone impedance)? *also assuming the amp's output impedance is negligible*
  4. castleofargh Contributor
    you're looking too much into it when you apparently have no idea how loud you might go when you're listening. so you're trying to be super precise starting from a value you don't have. using dynamic values of tracks that might not even represent how you listen. I also went with some matter of guessing, but not as much IMO(well at least I hope).
    I went at it the other way around. I used some measurements saying that the fiio X1 was pretty much steady in voltage from 32 to 600ohm and decided with no way to check it out that it would also be around that value into the 4000something ohm input in my laptop using a male to male jack cable. I felt safe using that as after all, as long as it's between 1 and 2V I would be wrong by less than 3db picking 1.5v as the value. but I can't really be sure the X1 doesn't strangely collapses with that load. it sounded fine but that's all I can say. the specs seem to show that it has more troubles with current into low loads than keeping it up in volts into high loads. I hope I was right.
    I used a MC5 for several days making sure to take note of the volume setting I used, I tried inside, outside, near cars(but they do isolate greatly so it doesn't change too much). and I ended up with a maximum volume setting. I then played a 1khz test tone(tried with some noise too), measured with a free RTA software the voltage/db variation between max output and the setting I had. still assuming I was in the 1.5V area maxed out on the X1 (a multimeter could be nice one day).
    now I know a voltage somehow and turn it into db with the MC5 specs.(the test tone was some db down, not recorded at 0db, so I compensated for it beforehand).
    I ended up just slightly above 105db with my mess. trying the same but going relatively loud on purpose(some loudness I could listen to but realy not for more than a song), I ended up at 114db, when the usual value suggested for amps is 115db... I'm half happy, half furious.
    add to that +/6db if the X1 is actually somewhere between 0.75v and 3V when maxed out into my laptop, I'm guessing it's a pretty safe bet and 120db should be the absolute worst case scenario ever using the most dynamic song, playing it the loudest, and making the most estimation mistakes.
    and more realistically, I would need 99 to 111db max.
    now the perk of my way, it's me using my DAP in real life with my real music and keeping the highest value I used. and what's best, the songs are not going above 0db anytime soon so I don't really have to care about the dynamic or how many db down the track is recorded.
    now the problem is to know if we would listen at the same loudness with different headphones of different signatures. you might need to take the FR into account a little. but apart from that, you end up with an estimated listening loudness so you can check the needs of any headphone.
    else I thought about estimating an average listening level one way or another, and just add 23db as that's the headroom for the EBU R128. it's the biggest headroom so far (it was like 14db for old replay gain) and I do feel like it's working great so that might be a way to rapidly get in the ballpark of your needs. with your measured RMS last page at a meeting reaching up to 80db, that's 103db peaks for normal listening, 113db for subjectively twice as loud. no drama and we always seem to end up in the same zone.
  5. Joe Bloggs Contributor

    ...Or you could learn to listen for the effects of distortion. True distortion is anything but a vague feeling of "this doesn't sound like it's being driven to its full potential..."

    Start with sine tones at different frequencies with near 0dBFS magnitude. Peg the volume of foobar and the system to max and adjust the output gain of foobar to positive until you hear clipping (turning down the volume at the DAC / amp of course).

    Using the pop-out peak meter (which displays above 0dBFS, unlike the toolbar peak meter, check whether the signal is above 0dBFS and if so how far above. If you start to hear distortion starting exactly above 0dBFS you're doing good. Otherwise, your DAC/amp combo does not play cleanly up to full digital scale and you may want to keep your digital volume below 0dBFS.

    Now try reducing system volume from max until you get comfortable travel on the volume knob of your DAC / amp and repeat the experiment. It should definitely only clip above 0dBFS on foobar now.

    Now try playing some familiar piece of music through foobar and repeat--adjusting the gain until you start hitting past 0dB at peaks. At what point do you hear obvious distortion? What does the "in-between zone" sound like? Does it correspond at all to those uneasy feelings of "not enough power" you were getting?
  6. Steve Eddy

    Sure. You'd just have to know how much attenuation the volume control was providing at each position, and what the gain of the amplifier is, and then it's just simple math.

    That would be pretty easy for linear attenuation (i.e. attenuation would be 50% or 6 dB at the middle position). But we perceive loudness somewhat logarithmically, so volume controls for audio typically aren't linear. Here's a graph for some audio taper volume controls. A linear taper would be a straight line from 0,0 to 100,100. But as you can see, a given rotation of the knob doesn't have a linear relationship to the amount of attenuation. So basically, you'd have to feed the amp a sine wave of a known voltage (though low enough in level that the amplifier doesn't clip when you have the volume turned all the way up, and then write down the output voltage for each position of the volume control.


  7. castleofargh Contributor

    I have a feeling that the post wasn't addressed to me ^_^.
  8. imackler
    Of course, I'm skeptical :) But did anyone get a chance to AB the Magni and Magni 2 and hear a difference in sound signature/quality? 
  9. maverickronin
    How exactly did this thread get locked before anyone even made a reply...
  10. Steve Eddy

    How do you know there weren't replies? :wink:

  11. maverickronin
    Did you make one?  [​IMG]
    I just assumed it would be easier to delete the whole thread instead of deleting all the replies and then locking it...
  12. Steve Eddy

    Nope. I saw it same as you. No replies.

  13. davidsh
  14. bigshot
    I answered the fella's questions in PM. He thanked me for replying. I don't think he knows why it was locked either.
  15. Steve Eddy

    Weird indeed.

100 101 102 103 104 105 106 107 108 109
111 112

Share This Page