[Tadgh]

I'm confused regarding headphone amplification, and could do with some totally objective and scientific clarification, if anybody could be as good as to enlighten me that'd be greatly appreciated
 
I have a fairly basic understanding of the science of sound (okay super basic). As I see it, electricity powers a driver at certain frequencies in order to produce certain tones. Drivers that require a lot of energy (that are inefficient) or use particularly resistive components (are high impedance), require more power. Cleaner amplification leads to cleaner signal, hence why the capacitors on an amplifier are crucial.
 
Where I get confused is:
 
 
Why amps have sonic charictaristics (and aren't all just mirror images of their input signal with a little added distortion)
 
Why amps have so many different metrics, and what these measurments mean
 
Why leaving amps to "warm up" supposedly offers an improvement in SQ
 
What makes a "good" amp
 
 
If any examples are used, please use these metrics (Just for my ease, I plan to buy said amp):
 
 
Frequency Response: 20Hz-20Khz, -0.1db, 2Hz-400KHz, -3dB
 
Maximum Power, 32 ohms: 1.0W RMS per channel
 
Maximum Power, 50 ohms: 1.0W RMS per channel
 
Maximum Power, 300 ohms: 380mW RMS per channel
 
Maximum Power, 600 ohms: 190mW RMS per channel
 
THD: Less than 0.008%, 20Hz-20KHz, at 1V RMS, high gain mode (worst case)
 
IMD: Less than 0.010%, CCIF at 1V RMS, high gain mode (worst case)
 
SNR: More than 102db, unweighted, referenced to 1V RMS, in low gain mode
 
Crosstalk: Less than -72dB, 20Hz-20KHz
 
Output Impedance: Less than 2 ohms
 
Gain: High = 6 (15.6db) or Low = 1.5 (3.5db), via rear switch

 

[MindsMirror]

 
Why amps have sonic charictaristics (and aren't all just mirror images of their input signal with a little added distortion)

You probably know that tube amps will add audible distortion and warmth to the audio, which is often intentional. Some amps will intentionally change the frequency response, like adding bass for example to appeal to particular people. Output impedance of the amp will change the frequency response of a dynamic driver headphone due to the headphone's impedance curve not being flat. The higher the output impedance, the more the frequency response is changed by the headphone's impedance curve. In many cases, amps do sound the same. You can find studies where people who believed that two amps sound different could not tell them apart in blind listening tests.
 
 

 
Why amps have so many different metrics, and what these measurments mean

No amp is perfect, the measurements are all describing how imperfect the amp is in various different aspects of it's performance. Are there any in particular that you are unclear on?
 
 

 
Why leaving amps to "warm up" supposedly offers an improvement in SQ

This applies only to tube amps. Vacuum tubes need to be warm in order to function. They have a heating element in them which can take a few minutes to reach a stable temperature.
 
 

 
What makes a "good" amp

Like I said above, the specs tell you in what ways the amp is imperfect. A perfect audio amp would have:
 

Frequency Response: 20Hz-20Khz, -0db
 
Maximum Power: As much as you need
 
THD: 0%
 
IMD: 0%
 
SNR: - infinity dB
 
Crosstalk: - infinity dB
 
Output Impedance: 0 ohms
 
Gain: As much as you need

This theoretical amp would be impossible to build, but as long as all of these parameters are below your audible threshold, it will sound perfect. You can't hear a 0.1dB variation in frequency response, or a -102dB noise floor, or 0.01% distortion, so the amp you plan to buy should be audibly transparent.

 

[Tadgh]

Spoiler: Warning: Spoiler!

  You probably know that tube amps will add audible distortion and warmth to the audio, which is often intentional. Some amps will intentionally change the frequency response, like adding bass for example to appeal to particular people. Output impedance of the amp will change the frequency response of a dynamic driver headphone due to the headphone's impedance curve not being flat. The higher the output impedance, the more the frequency response is changed by the headphone's impedance curve. In many cases, amps do sound the same. You can find studies where people who believed that two amps sound different could not tell them apart in blind listening tests.
 
 
No amp is perfect, the measurements are all describing how imperfect the amp is in various different aspects of it's performance. Are there any in particular that you are unclear on?
 
 
This applies only to tube amps. Vacuum tubes need to be warm in order to function. They have a heating element in them which can take a few minutes to reach a stable temperature.
 
 
Like I said above, the specs tell you in what ways the amp is imperfect. A perfect audio amp would have:
 
This theoretical amp would be impossible to build, but as long as all of these parameters are below your audible threshold, it will sound perfect. You can't hear a 0.1dB variation in frequency response, or a -102dB noise floor, or 0.01% distortion, so the amp you plan to buy should be audibly transparent.

 

Hey @MindsMirror!
 
So what makes tube amps curb higher frequencies? Right so, if one wanted to build an amp that added bass, would they reduce the relative impedence of bass frequencies? How do impedence curves in amps and headphones interact with one another?
 
Right, that clears up a lot (that theoretical perfect amp is really helpful). Regarding specific metrics:
 
How exactly does Gain work?
What is SPL?
What is SNR?
How does one relate amp output power and driver efficiency? (if I wanted to buy a pair of HD650s >300 Ohms and an SPL of 103db, how would I figure out if that amp I mentioned would adequately power them? I know it is, but how does that math work?)
 
Audibly transparent, that's high praise indeed They're the specs from the Asgard 2, I'm gonna take that as a potential SS end-game (certainly unless I get into electrostatics or something).
 
Thanks so much for your response, it's been very helpful

 

[MindsMirror]

 
So what makes tube amps curb higher frequencies? Right so, if one wanted to build an amp that added bass, would they reduce the relative impedence of bass frequencies? How do impedence curves in amps and headphones interact with one another?

Tubes don't perform as well as transistors, I guess they just have a less flat frequency response, so the treble rolls off sooner.
 
If you're designing an amp with a boosted bass, you would add a filter before the output stage, which would change the frequency response without affecting the output impedance.
 
In a perfect amp with no output impedance, all the voltage the amp produces will be dropped on the headphone. If your amp has some output impedance, some of the voltage it produces will be dropped internally, not on the headphone. If the headphone impedance varies at different frequencies, then the ratio of the voltage dropped on the headphone to the voltage dropped internal will be different at the different frequencies. That means that the frequency response of what the headphone receives is altered.
http://www.innerfidelity.com/images/SennheiserHD800.pdf
Here you can see a graph of the HD800's impedance. It has a peak at 100Hz. A higher output impedance in the amp will cause the frequency response to be changed in the shape of that curve, so for that headphone, an amp with high output impedance would have a peak in frequency response at 100Hz compared to an amp with zero output impedance.
 
 

 
How exactly does Gain work?
What is SPL?
What is SNR?

Gain is just the amount of amplification. If you have a gain of 6 and you input 1 volt, it will output 6 volts. If your had only a fixed gain there would be no way to adjust the volume, so there is also a volume knob which lowers the level either before or after the gain is applied.
 
Sound Pressure Level is just the way of measuring loudness of a sound. It's measured in dB relative to the quietest sound that humans can detect, which is defined as 0dB.
 
Signal to Noise Ratio is a measure of how quiet the background noise is. If your amp's SNR is 102dB, then it could play a sound that is 102dB loudness (very loud), and the background noise would be at 0dB (barely perceptible). The noise in a typical room is actually around 30dB, so your amp could actually play a sound at 132dB (deafening) before its noise floor could be heard over the noise in your room.
 
 

  How does one relate amp output power and driver efficiency? (if I wanted to buy a pair of HD650s >300 Ohms and an SPL of 103db, how would I figure out if that amp I mentioned would adequately power them? I know it is, but how does that math work?)

The sensitivity of 103dB SPL @ 1V means that the headphone will produce 103dB of sound if the amp sends it 1V.
 
The amp is capable of 380mW into 300 Ohms, and using Ohm's law (P = V^2/R) to convert that to voltage, it equates to 10.7V, roughly 10 times higher than 1V, meaning it can go 10 times louder than 103dB. 10 times louder equates to an increase of 20dB, so the actual SPL it could produce is 123dB.
 
Music has peaks above the average sound pressure level, and depending on how dynamic your music is you will need more or less headroom. I don't think even the most dynamic of music would go much over 23dB peaks (the R128 standard for loudness normalizes everything to -23dB, a value which was chosen because it has enough headroom for almost anything), so the actual average SPL the amp could produce with dynamic music would be about 100dB. You (hopefully) don't listen to anything that loud. Up to 90dB is the most I would expect anyone to actually listen to, and the amp is certainly capable of that with those headphones.
 
You can probably find calculators online where you can input the numbers and it will do all of those calculations for you.

 

[Tadgh]

  Tubes don't perform as well as transistors, I guess they just have a less flat frequency response, so the treble rolls off sooner.
 
If you're designing an amp with a boosted bass, you would add a filter before the output stage, which would change the frequency response without affecting the output impedance.
 
In a perfect amp with no output impedance, all the voltage the amp produces will be dropped on the headphone. If your amp has some output impedance, some of the voltage it produces will be dropped internally, not on the headphone. If the headphone impedance varies at different frequencies, then the ratio of the voltage dropped on the headphone to the voltage dropped internal will be different at the different frequencies. That means that the frequency response of what the headphone receives is altered.
http://www.innerfidelity.com/images/SennheiserHD800.pdf
Here you can see a graph of the HD800's impedance. It has a peak at 100Hz. A higher output impedance in the amp will cause the frequency response to be changed in the shape of that curve, so for that headphone, an amp with high output impedance would have a peak in frequency response at 100Hz compared to an amp with zero output impedance.
 
 
Gain is just the amount of amplification. If you have a gain of 6 and you input 1 volt, it will output 6 volts. If your had only a fixed gain there would be no way to adjust the volume, so there is also a volume knob which lowers the level either before or after the gain is applied.
 
Sound Pressure Level is just the way of measuring loudness of a sound. It's measured in dB relative to the quietest sound that humans can detect, which is defined as 0dB.
 
Signal to Noise Ratio is a measure of how quiet the background noise is. If your amp's SNR is 102dB, then it could play a sound that is 102dB loudness (very loud), and the background noise would be at 0dB (barely perceptible). The noise in a typical room is actually around 30dB, so your amp could actually play a sound at 132dB (deafening) before its noise floor could be heard over the noise in your room.
 
 
The sensitivity of 103dB SPL @ 1V means that the headphone will produce 103dB of sound if the amp sends it 1V.
 
The amp is capable of 380mW into 300 Ohms, and using Ohm's law (P = V^2/R) to convert that to voltage, it equates to 10.7V, roughly 10 times higher than 1V, meaning it can go 10 times louder than 103dB. 10 times louder equates to an increase of 20dB, so the actual SPL it could produce is 123dB.
 
Music has peaks above the average sound pressure level, and depending on how dynamic your music is you will need more or less headroom. I don't think even the most dynamic of music would go much over 23dB peaks (the R128 standard for loudness normalizes everything to -23dB, a value which was chosen because it has enough headroom for almost anything), so the actual average SPL the amp could produce with dynamic music would be about 100dB. You (hopefully) don't listen to anything that loud. Up to 90dB is the most I would expect anyone to actually listen to, and the amp is certainly capable of that with those headphones.
 
You can probably find calculators online where you can input the numbers and it will do all of those calculations for you.

Okay I think I understand to a decent degree now, thank you.

 

[madbass10]

Awesome post by both of you! I just joined and wanted to add some limited knowledge.
 
Quote:

 
This applies only to tube amps. Vacuum tubes need to be warm in order to function. They have a heating element in them which can take a few minutes to reach a stable temperature.

 
Tube amps certainly need to be warm to even function, but there is also a very minor effect in solid state amps. Cold devices behave differently than warm devices, especially semiconductors. Transistors can have much higher gain as their temperature increases, which is why there is almost always some local negative feedback to prevent thermal runaway. In most amplifiers, the difference is not dramatic, but it is measurable and sometimes noticeable. There may also be circuits in the amp like servos with very long time constants, that may take a while to stabilize.

 
 
So what makes tube amps curb higher frequencies? 

There's a lot of stuff cutting high frequencies in tube amps. The tubes themselves, being more or less a collection of conductive surfaces in close proximity to each other, are subject to a lot of parasitic capacitance that can cause problems if the frequency response is not limited. Tube amps also have output transformers, which limit bandwidth on the high end.

 

[arnyk]

 
There's a lot of stuff cutting high frequencies in tube amps. The tubes themselves, being more or less a collection of conductive surfaces in close proximity to each other, are subject to a lot of parasitic capacitance that can cause problems if the frequency response is not limited. Tube amps also have output transformers, which limit bandwidth on the high end.

 
The fact is that amps based on fairly ordinary tubes have been routinely used at RF frequencies up to 150 MHz and beyond. For example the 12AT7 which was widely used in audio was also widely used as the RF amplifier in FM tuners.  
 
It is true that tubed audio amps tended to have limited bandpass and many failed to be adequately flat up to 20 KHz. The most common reason was that tubed amps generally needed transformers to match the natural high impedance of the tubes to the lower impedance loads.