what makes some amp better than others?

Defining your use of "better" would help, but working with the general use of the word here is what I find in a good amp.

1.)  Low harmonic distortion ( THD < .002 % )

This gives a nice clean signal without distorting the input waveform

2.)  High signal to noise ratio ( SNR > 98dB )

This allows the noise floor to be low enough to prevent distortion and other adverse effects )

3.)  Larger frequency response ( FR: 10Hz - 30kHz )

At minimum you want to include the entire human audible spectrum which is typically 20Hz - 20kHz, but I find the wider the FR is the better.

4.)  Lower output impedance ( X < 5 Ohms )

Anything else higher can lead to, call it atypical, frequency response issues along with distortion

5.)  Output power at 1Khz ( P > 250mW )

This is something I look at to make sure the amp can adequately drive the many types of headphones out there.  Due to headphones having not only different impedances, but sensitivities as well, the higher the output power up to 1W the better.  Any higher is just ridiculous and too much power for most practical use headphones.

Those are the 5 things I look for in the specs in determining a better amp.  The next thing I do is check the physical design including how the power supply is implemented as well as functionality.  The final test is to listen to it and compare it against what I already have.

Thanks to nick for cathing my typo.

headphones require too big a range of signal for one amp to give good S/N with sensitive iem and drive inefficient orthodynamics to reasonable levels with any dynamic headroom

many amps recommended in head-fi threads are not suitable for the headphone - just a reflection of fanboyism, FOTM

any SS amp should have adequate frequency response, can have low output impedance - which can be "tuned" by adding series impedance  - tube amps often have higher output Z to start with - can't be made lower

but good  S/N, dynamic range can only be managed with amps designed for a range of headphone sensitivity, Z

amplifier with gain switching can help, but it would still be silly to use a 25 Vpk, 1/2 Apk output amp needed for an orthodynamic with sensitive iem that produce 120 dB SPL with <500 mV drive V

mis-matching headphone, amp drive requirements, abilities, gain means either hiss or clipping, transient distortion on dynamic music peaks

mostly imagination is the best scientific explanation - how many "comparisons" are done with any attempt at level matching, blinding, perceptual anchors, controls

absent noise, clipping or gross distortion levels the major determining factor in "sound" is frequency response - decent amps shouldn't have audible differences in frequency response

some do have inadequate signal coupling caps giving poor low frequency response, added series output impedance makes an amp sensitive to the headphones' impedance vs frequency variations

when these details are equalized even Stereophile's "Golden Ear" reviewers couldn't tell a $600 SS amp modded by Bob Carver from their own 4 figure US$ "SOTA" tube amp with their own choice of speaker, music

head-fi reviews, recommendations, "consensus" views of amp "sound" are better studied as social psychology - egged on by the Audiophile marketing press in ignorance, denial, and even active hostility towards the real results of psychoacoustics, engineering

Quote:

Originally Posted by Hughkk 

that's interesting,

u think that after certain price point, a SS amp and a tube amp will sound the same?

Price has nothing to do with quality, and so long as audio remains this unscientific and exclusive it never will.

A cheap tube amp designed for transparency (and that measures well) will be transparent. A five digit solid state amp designed to match the stereotypical "tube sound" will not be transparent.

So, let's take these ideas to their logical extreme. Consider the $20 Fiio E1, which is powered directly from an iPod as its power source. 

Tech specs here: http://www.fiio.com.cn/product/index.aspx?ID=14&MenuID=020301

Question: Should the E1 drive the Audeze LCD2s adequately? (tech specs here http://audeze.com/audeze-lcd2-bamboo) (Certain audiophiles might call me a blasphemer just for asking . . . )

If Fiio is to be believed, then the E1 does fine in criteria 1, 2, and 3 from the quote above (harmonic distortion, signal to noise ratio, and frequency response).

Fiio doesn't tell us anything about the output impedance . . . but then again neither do most amp manufacturers. 

As for power, the LCD2 has an efficiency of 91dB at 1mW and an impedance of 60 Ohms. Science tells me that power requirements multiply by 10 for every 10 decibel increase in sound, (i.e. if r = (decibels in excess of 91)/10, then power usage of LCD2 = 1 mW * 10^r ) and this is verified by the observation that the LCD 2 is rated for a maximum output of 133dB at 15 Watts. 133-91 = 42. 15 Watts = 1mW * 10^4.2 (approximately). A volume of 101dB is dangerously loud, so let's assume that we don't want to listen to music substantially louder than that. That would require a power usage of 1* 10^1 = 10 mW. The Fiio E1 is rated for 12 mW output at 300 ohm impedance and 100 mW output at 16 ohm impedance. The output at the 60 ohm impedance of the LCD 2 should be somewhere in between those, and at any rate, well above 10 mW.

So in conclusion, the Fiio E1 should drive the LCD2 just fine to volumes of over 100 dB, unless its unknown output impedance is too high or there is a problem with its physical design.

I really hope that's true, because I'd like to buy a pair of LCD 2s, and it would be great if I could amp them for $20. On the other hand, I have a suspicion that it's not true. The E1 is powered directly from an iPod -- it doesn't require any separate power source. Surely that should not be sufficient to power a massive headphone like the LCD2, at least according to popular wisdom. And yet the calculations in the previous paragraph show that it should, unless you want to go to volumes that would cause hearing loss very quickly. 

I can see two possible types of problems with the line of reasoning I've expounded above: 

1) NA Blur's criteria may be incomplete or incorrect, or I may be interpreting them incorrectly -- maybe the physical design part at the end is really important

2) Fiio or Audeze might be lying or mistaken about technical specifications

Anybody with a pair of LCD2s want to give it a try with the E1 and report back?

Or anyone want to point out what's wrong with my reasoning?

I am really puzzled by headphone amplifiers.

@Norman,

Take a look at my thread (link below), if you can find anything useful. 

lets add to my headwize link count: http://gilmore2.chem.northwestern.edu/articles/hearing_art.htm

you can look at a few of my explanations of how to use this info:

http://www.head-fi.org/search.php?advanced=1&search=hearing_art&titleonly=0&byuser=jcx&output=posts&replycompare=gt&numupdates=&sdate=0&newer=1&sort=relevance&order=descending&Search=SEARCH

basically it is "reasonable" to want Dynamic Headroom - ability to reproduce fractional second dynamic peaks such as snare drum hits, cymbal crashes at 120+ dB SPL

if you've spent  $k on headphones you are probably in the group of audiophiles that want this dynamic peak headroom for occasional live music SPL playback without worrying about rare clipping events on natural music transient peaks of well mastered music

100 dB clipping level is only adequate for heavily dynamic range compressed "Loudness War" victim music  - like all pop - and even then at much lower average SPL than typical live performance levels - yes live pop music is "unsafe" by OSHA standards

for iPod typical all day use you should be listening at <80 dB average level

it appears from the specs that the Fiio E1 must be using the TI TPA6130a2 - "Direct Path" chip that uses a internal V doubler to invert the 3.3 V power and give a bipolar +/-3 V drive with the headphones directly gnded - no output coupling C

the calculation of the power it can put out into 60 Ohms is a little complicated since you have to calculate an effective supply V and series R from the 2 differing impedance load, pwr ratings in the E1 spec

after a few simultaneous equations I get ~ 48 mW, ~= +17 dB re 1 mW into 60 Ohms

so the E1 should be able to drive the LCD-2 to 108 dB SPL peak before clipping - and in fact should be fine for all safe long term listening scenarios - anything but trying to reproduce live event SPL "Rock Out" levels of uncommonly well mastered high dynamic range music

Proton, thanks for the useful links and info. One of your links, http://sound.westhost.com/amp-sound.htm, explains that solid-state amps generally have output impedance below 1 ohm, so the output impedance should not be a worry when evaluating Fiio amps. The low output impedance allows an amp to control resonance more effectively, if I understand your explanation correctly. You also point out that impedance often varies with frequency. However, in the case of the LCD2 rev 2, the graphs at innerfidelity show that the impedance is constant with respect to frequency. 

Thanks, jcx, for the useful information. I had not understood dynamic headroom properly. I have read extensively about volume safety levels, but it's nice to see those links again.  I have measured decibel levels out of my headphones with an sound pressure meter, as described in this thread. http://www.head-fi.org/t/148786/measuring-decibels

I browsed through the threads that you linked. Apparently, the basic idea is that on well-mastered music, the instantaneous dynamic peaks can be 20-24 dB louder than the decibel level that shows up on my sound pressure meter on slow mode. When the meter is on fast mode, it still will not be fast enough to pick up the entirety of these peaks, especially since I have to turn a dial to move up to the next 10 dB range on the meter. These peaks are short-lasting enough that my hearing safety can be calculated without taking them into account, but reproducing them accurately will improve the listening experience.  

Could you explain the details of how you calculated the power output of the E1 at 60 ohms? I am a professional mathematician, so feel free to get technical. Also, how did you figure out which chip was used and the specs of that chip? I would like to make the same calculations for other portable amplifiers. 

I am the kind of person who is considering spending $1000 on a pair of headphones (haven't done it yet), but I'm not the kind of person who particularly enjoys listening to music locked in one place and hooked up to a heavy amp. I find that my enjoyment of listening to music on headphones is a function not only of the sound quality but also of my ability to move around while I listen to the music and by the extent to which I can easily operate the music reproduction system. The more complicated and bulky the amp, software, music player, etc. becomes, the less I can enjoy the music because the more I enter into an analytical mental state rather than an emotional "enjoy the music" mental state. Whereas when I can just pick up my iPod and listen to music quickly and conveniently, I can much more easily enter the mental state that allows me to enjoy music. So to hear that a simple device like the E1 could effectively drive the LCD2 as long as the volume not including peaks is kept under 90 dB is very good news to me. Indeed, for much of my listening, I listen well below 90 dB, say at around 70 dB. Sometimes, I do like to listen loud, even above 90dB, and for that I would need a better amp, depending on the level of the dynamic peaks in my music.

If I understand all this discussion right, though, then even with the most demanding headphones (not including electrostatics), if the volume is kept at very low but audible levels  (i.e. 55 dB average, with dynamic peaks maxing at 80 dB), then they can be driven just as well unamped from an iPod output as compared with a $10,000 amplifier. It is only when the volume is turned up that an amplifier provides any benefit. Yet I think many people on this forum would not agree with that claim. Is it safe to say that this is just a matter of psychology?

the Fiio E1 had enough clues in the product description - the 100 mW into 16 ohms, and the 0.009% distortion spec, volume control only fit one TI headphone op amp I could find

the actual math is middle school level once you have the equations - which come from simple electrical relations

to estimate 60 Ohm power capability from 100mW @16 Ohm and 12mW @300 Ohm rearrange Power = (V^2)/R to solve for the V across each load resistor

the system model I used was a unknown constant Vsource and a unknown Rseries common to both load conditions

then use the voltage divider relation to get 2 equations in the unknowns and the above calculated V across each of the 16, 300 Ohm load resistors from the power specification

V_16 = Vsource * ( 16 / (16 + Rseries) )...

solve the equations for the unknowns; I get Vsource ~=1.95 Vrms, Rsource ~= 8.70 Ohms

then use the V divider equation with 60 Ohms to find the V across a 60 Ohm load, calculate the power with this V

I use SciLab, a free MatLab workalike for calculations, Sage Math would probably let me skip the envelope back symbolic equation manipulation steps and do it all on the computer

        ___________________________________________       
                        scilab-5.3.3

                Consortium Scilab (DIGITEO)
              Copyright (c) 1989-2011 (INRIA)
              Copyright (c) 1989-2007 (ENPC)
        ___________________________________________       
 
 
Startup execution:
  loading initial environment

-->u_16=1.6^.5
 u_16  =
 
    1.2649111 
 
-->u_300=(.012*300)^.5
 u_300  =
 
    1.8973666 
 
-->z_0=(u_300-u_16)/(u_16/16-u_300/300)
 z_0  =
 
    8.6956522 
  
-->u_0=u_16*(16+z_0)/16
 u_0  =
 
    1.9523627 
 
-->u_60=u_0*(60/(60+z_0))
 u_60  =
 
    1.7052282 
 
-->p_60=u_60^2/60
 p_60  =
 
    0.0484634 
 
-->10*log10(p_60/0.001)
 ans  =
 
    16.854138 

(I cut out some intermediate, redundant steps, z_0 equation comes from the envelope back)

Okay, I understand how you did the calculations. Am I correct that the equation  V_16 = Vsource * ( 16 / (16 + Rseries) ) comes from the op amp specifications? How did you look up these specifications?

that's the equation for the division of the unknown Vsource with a Rseries connected to the 16 Ohm resistor:

http://en.wikipedia.org/wiki/Voltage_divider  Vin in fig 2 diagram = my Vsource, or u_0 in calcs; R1 = Rseries, z_0; R2 = 16 or 300 Ohm load as speced by E1; Vout = u_16 or u_300, or u_60 in my final calc

I didn't write out the equation for 300 ohms - just replace 16 with 300 everywhere, use the V calculated from the 12 mW into 300 Ohms

these 2 equations have the 2 unknowns Vsource, Rseries in common

the equations don't tell us where the Rseries is - whether it is a limitation of the op amp output stage, a added external series R or a limitation of the power supply

I suspect mostly the last - the power supply inverter using a switched C charge pump behaves approximately like it has a resistance limiting the current that is a function of the flying capacitor, switching frequency and any internal switching element resistance