[Xplo]

So I've been trying to get a feel for the differences between headphone amps, why one is better than another and so forth.. but most of the information I can find is either of the extremely basic "omg teh amp will give ur headphone more powar + maek everythng better! lol" variety or else shamelessly dives into the EE ocean with little regard to who drowns in it...

Is there something in the middle that explains things in more detail without getting *too* technical? For instance, I'd be curious why one opamp is preferred over another one given a particular design (since preferences seem to vary by amp design), or why some headphones need more voltage while others need more amperage. I might search Google except I have no idea what I'd be looking for, so instead, I hoped someone here could steer me right.

 

[john_jcb]

The DIY forum may be a better source of answers to your questions so I have taken the liberty of moving your thread there.

 

[lini]

xplo: Well, I won't comment on the amps, as the amp builders on the forum can do that better. For the headphones, it's a question of driver construction. Keep in mind that P = U x I and R = U / I, so P = U² / R: High impedance headphones will generally need more voltage, whereas low impedance headphones will need less voltage, because the lower resistance will allow a higher current flow. It's a bit like in cars: To achieve the same power, a big high torque engine needs fewer rpm than a smaller engine with lower torque.

Greetings from Munich!

Manfred / lini

 

[JohnFerrier]

Quote:

Originally posted by Xplo So I've been trying to get a feel for the differences between headphone amps, why one is better than another and so forth.. but most of the information I can find is either of the extremely basic "omg teh amp will give ur headphone more powar + maek everythng better! lol" variety or else shamelessly dives into the EE ocean with little regard to who drowns in it... Is there something in the middle that explains things in more detail without getting *too* technical? For instance, I'd be curious why one opamp is preferred over another one given a particular design (since preferences seem to vary by amp design), or why some headphones need more voltage while others need more amperage. I might search Google except I have no idea what I'd be looking for, so instead, I hoped someone here could steer me right.

Glad you asked. It's people's imaginations. Op amps are an order of magnitude, or better, lower distortion and more accuracy than any headphone can reproduce.

This is not too technical. Show me ANY op amp that produces 1khz waveforms like the Sennheiser HD-600 or the Grado RS-1 seen here:
http://www.headphone.com/graphCompar...are+Headphones

100hz:
http://www.headphone.com/graphCompar...are+Headphones

I don't know how Sennheiser can spec that at 0.1% THD 1khz. Looks closer to 10% to me. I doubt that our ears are sensitive enough to detect 0.1% distortion. (Especially when one considers that a 3db change--roughly what our ears can hear--is actually about a 50% change in signal.) Op amps/discrete amps are 0.01% or better.

Headphones and speakers are the weak link (by my estimatation an order of magnitude or more). It is not too hard to get a decent power supply and a decent source, so the next weakest link is the volume control (attenuator).

If you really want high fidelity, live music is the best way to go. Who knows what the THD is there. For rock, sometimes the higher the distortion the better. But at least you avoid all the less than state-of-the-art components used to record the music. And do you think that Diana Goldvocals or the band We Rock really care what resistors/capacitors were used to perform/record their music?

BTW: I am still attempting to constuct the best (lowest THD) head amp I can and hope to snag a Sennheiser HD-650 when they become available.

Just my thoughts.


JF

 

[JohnFerrier]

1) Try whatever headphone source you currently have (it may sound as good as 2, 3, 4, and 5).

2) Purchase a Creek OBH-21se ready to go for about $200.

3) Kevin Gilmore designed great DIY kits for $200/$400. Schematic/circuit description here:
http://headamp.com/order.shtml Kevin is doing a lot of things right with this design.

4) I keep going back to Borbely's design. I don't know if it can be beat. Great design and great parts for $600. Some think that Borbely has a large mark up, but if you look carefully at the level of design and the quality of components, it is hard really hard to beat. I continually find myself going back to it. Schematic/description here:
http://www.borbelyaudio.com/index41a.htm

5) If you really want to DIY, the best place to go is http://elvencraft.com/ppa/ Following the PPA group, you might just build the best headphone amplifier currently available.


JF

 

[tangent]

Quote:

omg teh amp will give ur headphone more powar + maek everythng better! lol

Great satire!

Quote:

I'd be curious why one opamp is preferred over another one given a particular design

Every active component (transistor or tube) has a different sonic character. This is a much more obvious thing than (alleged?) differences in totally passive components like resistors, wire, and connectors. I've never seen an explanation of how you can accurately predict the sonic character of a device just from looking at the specs. With op-amps (which contain lots of transistors) it's easy to just try them one after another and listen to the results, so people have, and there seems to be some broad consensus on what various op-amps sound like.

Now, the better op-amp based amps have some kind of buffering. I'm not just talking about popular DIY designs here: all of the Headroom amps and some of the Creek and Meier amps are buffered op-amp designs. Buffers are good at supplying lots of current without any voltage amplification. A buffer alone isn't useful because you need voltage amplification in most situations, and also buffers aren't accurate by themselves. (Lots of distortion.) If you wrap an op-amp around the buffer, the op-amp takes care of accurate voltage amplification and keeps the buffer accurate; the buffer in turn relieves the op-amp of having to drive the difficult load of the headphones by itself. Together, the combination sounds better than just an op-amp or a buffer alone.

I tell you all this because adding a buffer adds another active element in the signal path. Now you start getting into the mess of euphonic combinations. One op-amp might be aggressive sounding on its own, but if you follow it with a laid-back sounding buffer the combination might sound great. Put that same op-amp into an amplifier using aggressive buffers and the combination will be unlistenable.

As with all attempts at describing sound with words, this is going to come across as a lot of handwaving, I'm sure. All I can say is to ask you to try it for yourself if you're not willing to take my description on faith. I encourage you to do it.

Quote:

why some headphones need more voltage while others need more amperage.

Headpones have two main specs: sensitivity and impedance.

Sensitivity tells you how many milliwatts the headphone requires to reach a given volume level. A typical spec might be 82dB/mW, meaning that at one milliwatt, the headphones will put out 82dB. If you want to get to 90dB, that's an 8dB power difference which translates to about 6x the milliwatts. (Search on Google for "decibel power formula" if you want to know how I arrived at that number.)

Impedance is a complex topic, especially in regards to headphones. Instead, pretend that it's just resistance. Then you use Ohm's law: voltage equals current times resistance. This tells you that as resistance (impedance) goes down, voltage has to go up to keep current the same, or current has to go up to keep voltage the same.

Let's say you have some AKG K401s: the impedance is 120 Ohm and the sensitivity is 94dB/mW. Let's not adjust that any: we'll say our listening peaks are 94dB, so we need 1mW from our amp to reach that with these headphones. Then you use another formula (I=sqrt(P/R)) to find that you need about 3mA to put that much power into that much resistance. Going back to Ohm's law, voltage would be about 0.3V.

Say instead you have some Sony MDR-V700DJs. Impedance is 24 Ohm, sensitivity is 107dB/mW. Adjust down to 94dB (-13dB) and power is now 0.05mW. You need 1.4mA and 0.03V to put that much power into that load.

These are all idealized numbers. The sensitivity number just gets you into the right ballpark, and the impedance varies greatly with frequency. Still, I hope you now see how different headphones can require different voltage and current levels to get to the same volume level.

Quote:

Op amps are an order of magnitude, or better, lower distortion and more accuracy than any headphone can reproduce.

You sound like a man of science, so try this experient: build up a basic op-amp amplifier (CMoy type) and try an OPA2132 and an AD8620 in it. Buffer it, if you'd prefer. If you can't hear the difference between the two despite the fact that they both claim distortion down in the 0.00x% range in their datasheets, well congratulations, you will be saving a lot of money on parts this year.

You may be right that humans cannot hear 0.1% distortion, but it's also true that a simple THD test doesn't cover all of the distortion mechanisms in an amplifier.

 

[JohnFerrier]

Quote:

Originally posted by tangent so try this experient...

This is what I plan to do. Build a PPA like amplifier: OPA637s with parallel OPA627s as buffers, 0.01% bulk metal foil resistors, bulk metal potentiometers (with a scheme to balance them to <0.1%--yes, they will be linear taper, turn with a small screwdriver type), teflon and air gaps between circuit layers, OFC silver plated wire/silver laced solder, spend $50 on 9V batteries (rather than even use a toroidal transformer), still use low noise regulators and be through with it. If I'm lucky, it will sound better than the headphone jack on my Yamaha. My contention is that the rather large distortion of the headphones themselves masks almost all differences in op-amps. (I don't even want to mess around with IC sockets to A/B op-amps.) I really think that advances in electronics has progressed past the point of perceptability (remember our hearing is logarithmic--whereas oscilloscopes and meters are capable of measurements well beyond our hearing ability), but headphones and speakers still lag *far* behind. Even the manufactures publish a relatively large 0.1% THD spec. (Also, note that it is usually spec'ed only at one point--1khz. There is no THD versus frequency graphs. Ever wonder why?) And being that the headphones are the weakest link, I'm very interested in the upcoming HD-650s (any new info about these appreciated).

A side question, about the PPA. If one uses a current source on the output buffers (say 50 mA for 300 ohm headphones) is the current source for the op-amp still necessary? To ask this another way. Does a CCS on the output drive both the buffer and the op-amp into class-A? My guess is that it would. (To be frank, I don't plan to do this--obviously, a lot of current draw for batteries. Just wonder if someone knows something about this.)


JF

 

[JohnFerrier]

If you listen to a 1khz sine wave in your HD-600 or Grado RS-1, are your ears sensitive enough to tell that the waveform is like the following: http://www.headphone.com/graphCompar...are+Headphones

Ghastly. And what do you think headphones do to complex wave forms??? I'd say our ears are not super sensitive...

I'm sure most of you have seen this. Just think about this though. Other headphones here: http://www.headphone.com/layout.php?topicID=10 I'd say that headphones dominate the sonic signature. And there is a technical reason. A headphone/speaker is physically much different than volin strings, Stratocaster strings, vocal chords, etc.--the sounds that they are attempting to recreate. In the chain of sound reproduction, the speaker has by far the greatest challenge--transform electricity into the sound of a full orchestra. The electronics between the microphone and the speaker are closing in on perfection. The two end points have a considerable ways to go. (If it were easier, the HD-650s would be released by now.)


JF

 

[lini]

tangent: Sorry, but sound pressure per power (dB/mW for headphones) is efficiency, not sensitivity (it's a common confusion, though). To be excact, sensitivity is sound pressure per voltage (dB/mV). Thus one has to be careful with these specs: E.g., sensitivity is generally high for low impedance headphones - but that doesn't necessarily mean that a particular low impedance headphone is easy to drive (= has good efficiency, too).

Greetings from Munich!

Manfred / lini

 

[morsel]

"This is what I plan to do. Build a PPA like amplifier: OPA637s with parallel OPA627s as buffers..."

John, opamps do not make good buffers. They also don't parallel well, as output resistors are required, which raises output impedance, and they have very low output current, generally in the 30-50mA range. You should consider using real open loop buffers such as as the HA5002 or BUF634.

The experimental amp you describe sounds more like a CHA47 than a PPA. You seem to want to cut many corners and rely on your personal belief that headphones will mask opamp imperfections. Very low output impedance and high current capability will result in greater control of the amplifier over the headphone drivers. If you are dead set on using opamps for your output stage, check out this opamp power amp from Meier Audio, which uses massively parallel opamp arrays to provide high current outputs:

 

[morsel]

To answer your side question, no, a current source on the output buffers of a PPA would not bias the opamp into class A. That requires its own current source.

 

[JohnFerrier]

Quote:

Originally posted by morsel The experimental amp you describe sounds more like a CHA47 than a PPA.[/IMG]

Wow, Morsel. I hoped that this forum was tolerant of a variety of viewpoints--even my crazy ones. Knowing that Sennheiser only need about 30mA current max to be extermely loud, I think four op-amps would be enough. Hadn't thought about using 44 (that is more hole than I plan to drill though).

Thanks for the information! I hadn't stumbled into Meier's design (and I've been to his website a few times).


JF

 

[morsel]

John, to get great sound from headphones you should be shooting for 1A, not 100mA, of output current. This is not about providing enough current, it is about having low enough output impedance to exert tight control over the drivers. 4 opamps are not enough. In order to parallel them you will need output resistors, which will raise the output impedance. If I were going the opamp route, I would probably try about 16 opamps in parallel. If you are going to be using a 2 stage topology anyway, why not do it right and use some decent open loop buffers instead? As an added bonus, HA5002 buffers are much cheaper than high end opamps.

Oops, I forgot you are using batteries. I'd still use open loop buffers, though.

 

[JohnFerrier]

Quote:

Originally posted by morsel John, to get great sound from headphones you should be shooting for 1A, not 100mA, of output current. This is not about providing enough current, it is about having low enough output impedance to exert tight control over the drivers. 4 opamps are not enough. In order to parallel them you will need output resistors, which will raise the output impedance. If I were going the opamp route, I would probably try about 16 opamps in parallel. If you are going to be using a 2 stage topology anyway, why not do it right and use some decent open loop buffers instead? As an added bonus, HA5002 buffers are much cheaper than high end opamps. Oops, I forgot you are going with batteries. I'd still use open loop buffers, though.

You certainly have me reviewing my design ideas. I really like how the BUF634s heatsink to the PCB. People complain about the output protection, though I think the HA5002 has protection as well.

Another idea I've thought about is the crossfed differential driver seen on the last page of the AD8610 datasheet. I would have to purchase a spare Sennheiser cable and modify it to be differential. Maybe I'll go back to that idea.

I don't mind using the spec sheets as a guide (rather than what people hear). Spec-wise the OPA627/637 are better in several categories: THD, CMRR, and PSR. Also, for a gain of 10 (and probably over 5), the OPA637 is better than the OPA627. This is true for THD over 1khz (if you look carefully at the differently scaled datasheets).

Again, it is good to bounce thoughts with someone.

I assumed that the CHA47 was the Meier design. Is it a Headwise project? I can't find a schematic.


JF

 

[Joe Bloggs]

Quote:

http://www.headphone.com/graphCompa...pare+Headphones

I don't think you can just look at a time domain graph and tell how much harmonic distortion there is?? What you see is mostly the result of frequency response and phase shift, both of which are not harmonic distortion AFAIK??