Head-Fi.org › Forums › Equipment Forums › Sound Science › 24bit vs 16bit, the myth exploded!
New Posts  All Forums:Forum Nav:

24bit vs 16bit, the myth exploded! - Page 104

post #1546 of 1937
Quote:
Originally Posted by SilverEars View Post

This of course depends on the sensitivity of the headphones.  Even at 600ohms, if the sensitivity is high enough, it will be loud at 1.5 or 2 volts. In the case of Beyers, it's not the case so yeah, it need higher voltage, so the puny DAPs are not providing enough juice.

I think people generally think low impedance as easier to drive because lots of portable headphones are low impedance and they get loud, not because of the low impedance, but because of the high sensitivity.  

This is why HE-6 at 50ohms is still hard to drive because the sensitivity is low, which is quite different from most headphones specs.  That probably throws people off.

While sensitivity plays a role ( e.g. He-6 are like 50 ohms, but terribly inefficient) physics works against high impedance drivers being loud, given a fixed voltage and driver efficiency.

The electrical power available to convert into acoustic power for a voltage amplitude of V and a speaker impedance R is V^2/R. Therefore the the power decreases with increasing impedance.

Cheer
post #1547 of 1937
Quote:
Originally Posted by Dark_wizzie View Post
 

If higher ohms mean easier to drive, then why can't all amps just drive 600 ohms?

Most well designed desktop amps should be able to drive 600 ohms. As ab initio said though, usually, with 600 ohm headphones, an amp will run out of voltage before it runs out of current, so the limiting factor tends to be the output op amp (or discrete amp, if the designer so chooses), and the voltage supplied to that op amp by the power supply (you need a higher supply voltage to have a higher output, unsurprisingly). Most mobile devices can only go to 1 or 2 volts RMS before running out of voltage, usually because they are driven off a fairly low voltage battery, but any decent desktop design should be able to hit the 7-8V RMS needed to drive some of the less sensitive 600 ohm models.

post #1548 of 1937
Quote:
Originally Posted by SilverEars View Post
 

Well, I have a tube hybrid amp I use with the 650 and when I used it with the 800 it started to crackle at a certain volume level, so my guess was that it was starting to clip.  I had no problems like that with my 650.  I've also tried the 800 with my O2, and I recall it sounded worse than my 650(don't recall if it clipped or not), but I don't like listening to my 650 with the O2.  

 

I've tried O2 and my Beta 22 with LCD2 and HE-6.  There is noticeable difference in the sound, and I favor the Beta 22 as both of them sound tighter and has better dynamics. 

Hmm. Given that it's a tube hybrid amp, it probably wasn't designed with low distortion and accuracy in mind, but I wouldn't have guessed that it would crackle or have any issues with the 800 if it plays the 650 just fine. The 800 is higher impedance at all frequencies (which, as I said, makes it easier on the amp), and it is only 1dB less sensitive at the same voltage (102dB at 1V, vs 103dB at 1V) by Sennheiser's spec. I don't know if there's much more to be said on the subject without measurements of the amp playing into each headphone though.

 

As for the O2 vs Beta22, have you done a level-matched blind test? I haven't seen a detailed measurement of the B22 from an independent source, but I would expect it to measure near perfect based on AMB's spec, so I would expect both it and the O2 to sound identical with the LCD2. With the HE-6, I would expect them to sound identical at low to moderate volume, but at very high listening levels, the O2 may hit its current limiter, which would be very audible (and make it sound quite different from the B22). 

post #1549 of 1937
Quote:
Originally Posted by ab initio View Post


While sensitivity plays a role ( e.g. He-6 are like 50 ohms, but terribly inefficient) physics works against high impedance drivers being loud, given a fixed voltage and driver efficiency.

The electrical power available to convert into acoustic power for a voltage amplitude of V and a speaker impedance R is V^2/R. Therefore the the power decreases with increasing impedance.

Cheer

Sure, but increasing the voltage swing of an amp is pretty straightforward within the range that headphones require - it's often as simple as just increasing the supply voltage (and possibly heatsinking a couple parts). Most amplifiers have lower distortion into higher impedance too, even at a higher voltage level. High current with low distortion is more difficult to achieve, and for a fixed driver efficiency, the power input to a speaker is I^2*R, so the higher the impedance, the lower the current required for a given power level. This is why a low impedance, low efficiency speaker/headphone is so hard to drive.

 

For some numerical examples, to drive an HD800 to 110dB, you will need about 2.6V RMS, and (depending on frequency) around 4-9mA of current, which is 10-20 milliwatts of power. To drive a Beyerdynamic T1 to the same level requires just 2V RMS and only 1.5-3.5 mA, which is only 3-7mW (again, depending on frequency). So, even though the Beyer is higher impedance, it actually takes a bit less voltage and less than half the power and current to drive it to the same level. This means that the HD800 is harder to drive in every sense than the T1, and any amp that can drive the 800 should comfortably be able to drive the T1. Now, to look at the other end of the impedance spectrum, an Audeze LCD-2 would require 1.8V RMS to hit 110dB (nearly as much voltage as the Beyers) and around 25mA of current, resulting in 47mW of power. So, an amp for the LCD-2 would need just about as much voltage swing as an amp for the T1, but ten times the current, making them harder to drive. Compared to the HD800, the LCD-2 requires almost as much voltage, but around 2-3 times the current, so again, I would consider the LCD-2 the harder headphone to drive between the two (though there could potentially be some corner cases where an amplifier can just barely supply enough voltage for the LCD-2, but has plenty of current capability. This hypothetical amp would drive the LCD-2 just fine, but clip on loud passages with the HD800). Finally, let's look at the HE-6. With a 50 ohm impedance and an 83.5dB/1mW sensitivity (among the lowest I've ever seen for a headphone), it requires 5V RMS to hit 110dB, and in the process, pulls 100mA and about half a watt of power. For the same sound pressure level, it takes twice the voltage swing of an HD800, 10-20x the current, and 25-50x the power. This is why many headphone amps would have trouble with these headphones.

 

Interestingly, the O2 should actually be able to achieve this 110dB SPL with the HE-6, since at 50 ohms, it can achieve a voltage swing of about 5.5-6V. This doesn't leave a lot of headroom over the 5V required for 110dB though, and if you are listening to very dynamic music, this could still be inadequate. That's why I consider the O2 to be borderline for the HE-6. All of the other numbers I calculated above though are easily within the O2s capability with a huge amount of headroom (and within the capability of many other amps - 2-3V RMS and <50mA of current should be within the abilities of nearly every headphone amp on the market).


Edited by cjl - 5/9/14 at 9:03am
post #1550 of 1937
Quote:
Originally Posted by cjl View Post

 

I agree you.  Interesting thing about my HE-6 is that I can drive it to loud volume with my O2.  Also, I've been reading about variance to HE-6 impedance by various people.  Some people say they measure 40ohm and others say 60.  Not sure if HM goofed and labled HE-500 as HE-6, but I thought it was quite interesting.


Edited by SilverEars - 5/9/14 at 9:16am
post #1551 of 1937
Quote:
Originally Posted by cjl View Post
 

Sure, but increasing the voltage swing of an amp is pretty straightforward within the range that headphones require - it's often as simple as just increasing the supply voltage (and possibly heatsinking a couple parts). Most amplifiers have lower distortion into higher impedance too, even at a higher voltage level. High current with low distortion is more difficult to achieve, and for a fixed driver efficiency, the power input to a speaker is I^2*R, so the higher the impedance, the lower the current required for a given power level. This is why a low impedance, low efficiency speaker/headphone is so hard to drive.

 

For some numerical examples, to drive an HD800 to 110dB, you will need about 2.6V RMS, and (depending on frequency) around 4-9mA of current, which is 10-20 milliwatts of power. To drive a Beyerdynamic T1 to the same level requires just 2V RMS and only 1.5-3.5 mA, which is only 3-7mW (again, depending on frequency). So, even though the Beyer is higher impedance, it actually takes a bit less voltage and less than half the power and current to drive it to the same level. This means that the HD800 is harder to drive in every sense than the T1, and any amp that can drive the 800 should comfortably be able to drive the T1. Now, to look at the other end of the impedance spectrum, an Audeze LCD-2 would require 1.8V RMS to hit 110dB (nearly as much voltage as the Beyers) and around 25mA of current, resulting in 47mW of power. So, an amp for the LCD-2 would need just about as much voltage swing as an amp for the T1, but ten times the current, making them harder to drive. Compared to the HD800, the LCD-2 requires almost as much voltage, but around 2-3 times the current, so again, I would consider the LCD-2 the harder headphone to drive between the two (though there could potentially be some corner cases where an amplifier can just barely supply enough voltage for the LCD-2, but has plenty of current capability. This hypothetical amp would drive the LCD-2 just fine, but clip on loud passages with the HD800). Finally, let's look at the HE-6. With a 50 ohm impedance and an 83.5dB/1mW sensitivity (among the lowest I've ever seen for a headphone), it requires 5V RMS to hit 110dB, and in the process, pulls 100mA and about half a watt of power. For the same sound pressure level, it takes twice the voltage swing of an HD800, 10-20x the current, and 25-50x the power. This is why many headphone amps would have trouble with these headphones.

 

Interestingly, the O2 should actually be able to achieve this 110dB SPL with the HE-6, since at 50 ohms, it can achieve a voltage swing of about 5.5-6V. This doesn't leave a lot of headroom over the 5V required for 110dB though, and if you are listening to very dynamic music, this could still be inadequate. That's why I consider the O2 to be borderline for the HE-6. All of the other numbers I calculated above though are easily within the O2s capability with a huge amount of headroom (and within the capability of many other amps - 2-3V RMS and <50mA of current should be within the abilities of nearly every headphone amp on the market).

Can you do a for dummies version of your HD800 calculation? How did you get 2.6vRMS, is that the average case or the worst case scenario? So if we want to calculate the worst case scenario for HD800/O2 instead we should look at minimum impedance because that is hardest to drive? Then that would be 350 impedance which is a larger number than the 300 ohms listed on the headphones itself.

 

So you believe O2 will easily drive HD800, T1, LCD2/3.

 

And, higher impedance = requires more voltage, less current. Lower impedance = less voltage, more current. So in a way, having either a very high impedance or very low impedance can be an issue for an amp.


Edited by Dark_wizzie - 5/9/14 at 9:24am
post #1552 of 1937
Quote:
Originally Posted by Dark_wizzie View Post
 

Can you do a for dummies version of your HD800 calculation? How did you get 2.6vRMS, is that the average case or the worst case scenario? So if we want to calculate the worst case scenario for HD800/O2 instead we should look at minimum impedance because that is hardest to drive? Then that would be 350 impedance which is a larger number than the 300 ohms listed on the headphones itself.

 

So you believe O2 will easily drive HD800, T1, LCD2/3.

 

And, higher impedance = requires more voltage, less current. Lower impedance = less voltage, more current. So in a way, having either a very high impedance or very low impedance can be an issue for an amp.

 

So, with Sennheiser headphones, it's pretty easy. They specify their sensitivity at 1V RMS, so you don't actually have to worry about their impedance at all (unless you want to calculate power required or current). The HD800 are rated at 102dB at 1VRMS. From there, you can calculate the voltage needed for 110dB as shown here:

 

1) How much gain do we need compared to the reference? We want 110dB, reference specified is 102dB, so we want 8dB more SPL

2) For 8dB gain, 8 = 20*log(V2/V1)

3) V1 = 1 volt (specified reference level)

4) Rearranging the equation, 8/20 = log(V2)

5) 10^(8/20) = 10^log(V2) = V2

6) V2= 2.512

 

If you want to calculate the current or power requirements, you can then look up the impedance at the desired frequency, and use P = V^2/R and V = I*R to figure them out. If you want the worst case power and current, use the minimum impedance to do these calculations.

 

If you had a pair that specified in dB/mW, you would have an extra step involved. Let's try it with the HE-6:

 

1) How much gain do we need? Desired = 110, reference = 83.5, gain = 26.5dB

2) 26.5 = 10*log(P2/P1)    <----- Note the change in the leading constant - for voltage, use 20*log(V2/V1), for power, use 10*log(P2/P1)

3) P1 = 0.001W (1mW)

4) 26.5/10 = log(P2/0.001)

5)10^(2.65) = 10^log(1000*P2) = 1000*P2

6) 447 = 1000*P2 ---> P2 = 0.447W

7) P = V^2/R, R = 50, P = 0.447

8) V = sqrt(0.447*50)

9) V = 4.73

 

(and yes, I believe the O2 will easily drive the T1, HD800, and any of the LCD series)

 

As for your last statement, that is true - either very high or very low impedance can mean that a given amp will be unable to drive the headphones, and in either case, a very low sensitivity will increase the difficulty of driving the headphones.


Edited by cjl - 5/9/14 at 9:54am
post #1553 of 1937
Quote:

Originally Posted by Dark_wizzie View Post

Can you do a for dummies version of your HD800 calculation? How did you get 2.6vRMS, is that the average case or the worst case scenario? So if we want to calculate the worst case scenario for HD800/O2 instead we should look at minimum impedance because that is hardest to drive? Then that would be 350 impedance which is a larger number than the 300 ohms listed on the headphones itself.

 

So you believe O2 will easily drive HD800, T1, LCD2/3.

 

And, higher impedance = requires more voltage, less current. Lower impedance = less voltage, more current. So in a way, having either a very high impedance or very low impedance can be an issue for an amp.

Voltage sensitivity is taken at some reference frequency, typically 1khz. The frequency response chart shows you the output of the headphone though a range frequencies given a constant voltage. The current draw can be concluded from the impedance graph of the headphone.

 

The O2 should be able to power those headphones to near defeaning levels.

 

 

Interestingly enough, the HE-6 as measured by Tyll is far less sensitive, with an output of 77dB/mW and a 90dB output at 1 volt.

post #1554 of 1937
Quote:
Originally Posted by briskly View Post

Interestingly enough, the HE-6 as measured by Tyll is far less sensitive, with an output of 77dB/mW and a 90dB output at 1 volt.

90dB at 1V? That's even worse than my numbers above imply - that means to hit 110dB, you need 10Vrms and nearly 2 watts, well beyond the capability of the O2 (and the majority of other headphone amps out there).

post #1555 of 1937

There is a huge difference in the detail of transients between 16bit and 24bit.  I can absolutely hear it with my own ears.  Mixes never sound as good when I bounce them to cd, and it's not because I'm playing from a different source.  The files themselves sound less alive and bouncey in 16bit and when I'm mixing in 24bit.  I used to agree with much of what was stated in the above analysis, but that was before I had enough ear time with the differences to really tell. 

 

I'm sure I'm not alone in my sentiment.  There is also a very subtle difference between 24bit 48khz and 96khz perceptibly speaking.  I can't any tell any difference difference between 96 and 192.

post #1556 of 1937
Quote:
Originally Posted by warmer View Post
 

There is a huge difference in the detail of transients between 16bit and 24bit.  I can absolutely hear it with my own ears.  Mixes never sound as good when I bounce them to cd, and it's not because I'm playing from a different source.  The files themselves sound less alive and bouncey in 16bit and when I'm mixing in 24bit.  I used to agree with much of what was stated in the above analysis, but that was before I had enough ear time with the differences to really tell. 

 

I'm sure I'm not alone in my sentiment.  There is also a very subtle difference between 24bit 48khz and 96khz perceptibly speaking.  I can't any tell any difference difference between 96 and 192.

Ever done a proper double blind ABX?

post #1557 of 1937

*delete*

post #1558 of 1937
Quote:
Originally Posted by cjl View Post
 

90dB at 1V? That's even worse than my numbers above imply - that means to hit 110dB, you need 10Vrms and nearly 2 watts, well beyond the capability of the O2 (and the majority of other headphone amps out there).

With something like that, I'd just skip HE2 altogether. There are other good options that do not require this type of amplification.

post #1559 of 1937
Although I do agree with the article about the 24bit myth, 24bit/96Hz do have some uses for some listeners, that is, if they were to use some digital plugins like EQ, pitch change, tempo change, noise reduction etc.
Quote:
Originally Posted by warmer View Post

There is a huge difference in the detail of transients between 16bit and 24bit.  I can absolutely hear it with my own ears.  Mixes never sound as good when I bounce them to cd, and it's not because I'm playing from a different source.  The files themselves sound less alive and bouncey in 16bit and when I'm mixing in 24bit.  I used to agree with much of what was stated in the above analysis, but that was before I had enough ear time with the differences to really tell. 

I'm sure I'm not alone in my sentiment.  There is also a very subtle difference between 24bit 48khz and 96khz perceptibly speaking.  I can't any tell any difference difference between 96 and 192.
post #1560 of 1937
Quote:
Originally Posted by cjl View Post
 

 

So, with Sennheiser headphones, it's pretty easy. They specify their sensitivity at 1V RMS, so you don't actually have to worry about their impedance at all (unless you want to calculate power required or current). The HD800 are rated at 102dB at 1VRMS. From there, you can calculate the voltage needed for 110dB as shown here:

 

1) How much gain do we need compared to the reference? We want 110dB, reference specified is 102dB, so we want 8dB more SPL

2) For 8dB gain, 8 = 20*log(V2/V1)

3) V1 = 1 volt (specified reference level)

4) Rearranging the equation, 8/20 = log(V2)

5) 10^(8/20) = 10^log(V2) = V2

6) V2= 2.512

 

If you want to calculate the current or power requirements, you can then look up the impedance at the desired frequency, and use P = V^2/R and V = I*R to figure them out. If you want the worst case power and current, use the minimum impedance to do these calculations.

 

If you had a pair that specified in dB/mW, you would have an extra step involved. Let's try it with the HE-6:

 

1) How much gain do we need? Desired = 110, reference = 83.5, gain = 26.5dB

2) 26.5 = 10*log(P2/P1)    <----- Note the change in the leading constant - for voltage, use 20*log(V2/V1), for power, use 10*log(P2/P1)

3) P1 = 0.001W (1mW)

4) 26.5/10 = log(P2/0.001)

5)10^(2.65) = 10^log(1000*P2) = 1000*P2

6) 447 = 1000*P2 ---> P2 = 0.447W

7) P = V^2/R, R = 50, P = 0.447

8) V = sqrt(0.447*50)

9) V = 4.73

 

(and yes, I believe the O2 will easily drive the T1, HD800, and any of the LCD series)

 

As for your last statement, that is true - either very high or very low impedance can mean that a given amp will be unable to drive the headphones, and in either case, a very low sensitivity will increase the difficulty of driving the headphones.

 

Just want to make a quick comment here difference RMS and peaks (voltage or currents or power). Some care must be taken when talking about specifying these numbers. Manufacturers are not always clear on this.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › 24bit vs 16bit, the myth exploded!