[goodsound]

I am trying to understand the buffer concept in headphone amps.

From what I understand so far, a buffer has high input impedance and low output impedance so it will load the source 'lightly' and can handle high loads a its output i.e. is able to provide more current to its load.

Now, almost any device that you'd connect a headphone amp to (cdp, pcdp, pdp, soundcard) will already have some sort of an output op-amp. Ofcourse with very meagre current output capability and highish output impedance - more suitable to drive a high impedance load. But all the headphone amp designs that I saw here from cmoy to ppav2 have an "amplifier" (op-amp) also. Why is the amplifier op-amp needed ? Why not just a buffer that constitutes the "headphone amp" ?

 

[aerius]

Because people like to listen too darn loud, or because they have high impedance headphones where the output voltage swing of the CD player isn't enough to drive them to a decent volume. Or the source has a weak ouput. Or a combination of the above. Thus an op-amp or other such voltage gain stage is required.

 

[JahJahBinks]

Line out from CDP is not loud enough.

Also line out has no volume control, you don't want to have an amp that has only one volume setting.

 

[JaZZ]

The common 2 volts are enough even for high volumes, so a simple buffer stage would do the trick with every CDP and DAC. But like in speaker systems where active preamps are the norm, although all of the gain they provide has to be destroyed in the potentiometers again, headphone amps are an established item. One decisive argument in favor of them is that they allow to taylor the sound (by introducing active coloration), which isn't the case with buffers.
.

 

[00940]

Actually, a simple buffer with a pot before it is all you need for most cans (certainly not for my K340 though).

But... buffers color the sound too you know

A unity gain mosfet in class-A (such as the szekeres) and a discrete diamond buffer in class AB don't sound the same; neither do the buf634 or the ha5002. The problem is that most simple buffers have high(ish) offset if they're not in the feedback loop of an amplifier. They thus require an output capacitor or additionnal circuitry to stabilize them. It's thus just as easy to design a complete amp, even if you set it afterwards to unity gain.

 

[goodsound]

Ah…I forgot about the fact that a buffer is a unity gain amplifier. There is no voltage amplification so you are stuck with whatever max voltage the source can provide.

However, for all practical purposes, virtually any device that you could potentially connect your headphone amp to, would have a volume control and should be able to provide atleast a volt or two as maximum voltage. I’d imagine 1-2 volts across a typical 32 ohm headphone would be - loud. For connections without vol control like a LineOut - you could use a switchable voltage divider if you think a pot does not belong in the signal path.

But if it suits your purpose/application there is nothing technically wrong with the idea of using only a buffer as a headphone amp, right ? That’s what I am trying to find out.

P.S:- 00940, I saw your post only after I wrote the above. While searching on the subject of opamp buffers I did bump into what you mentioned - about buffers not in the feedback loop of the amplifier. Can you or anyone suggest some good reading(books, links,..) on OpAmps ?

 

[00940]

Mancini : Op amps for everyone : http://focus.ti.com/lit/an/slod006b/slod006b.pdf

Jung : Op Amp Applications Handbook : http://www.analog.com/library/analog..._handbook.html

It doesn't get any better than those two and they are free as pdf

 

[JaZZ]

Quote:

Originally Posted by 00940 But... buffers color the sound too you know

No, I wasn't aware of that -- so I stand corrected. But clearly less so than a complete amplification stage, I guess?
.

 

[goodsound]

Actually - as I think through it - wouldn't a simple, single stage common collector (emitter follower), which is nothing but a classic current amplifier, be the perfect headphone amplifier ?
Do such simple discrete headphone amplifiers exist ?

 

[aerius]

Yes, but they're not commercially available as far as I know. You have to buy the parts and build it yourself.

 

[goodsound]

a diy design would be great! can you point me to some ?

 

[aerius]

There's a lot of various designs over at the headwize project library, including a simple MOSFET driver.

 

[3DCadman]

What about just putting a pot in line with the line out, like a passive preamp does? Would you get enough gain from say, an Ipod?

I'm no EE, that was just a thought.

 

[JaZZ]

Quote:

Originally Posted by 3DCadman What about just putting a pot in line with the line out, like a passive preamp does? Would you get enough gain from say, an Ipod?

Not sure in the case of the iPod. There are two requirements: low output impedance and DC coupling or high (enough) serial capacitance. Some CDPs and DACs fulfill these requirements and can be used for driving headphones directly (preferrably high-impedance cans such as the HD 650). My Bel Canto works and sounds great this way, with a 500-ohm potentiometer as attenuator in the signal path.
.

 

[Solude]

Quote:

Originally Posted by JahJahBinks Line out from CDP is not loud enough.

Is that right? Huh cause my HeadAmp Reference is set to unity gain, meaning I can't go higher than the line out, and I still only listen at 10 o'clock position. 2v is very loud for headphones, on the Senn 650 thats 105?dB which is well into permanent damage, painful to listen to levels.