What about that buffer?

[ikoflexer]

I've been struggling with a headphone amp and being a n00b I came to the realization that I need a good buffer. Maybe not need, but it'd be nice to have a good buffer. Wishful thinking:

* seeeeeeeeeempuuuuuul
* low distortion into a typical low impedance headphones load
* high input impedance
* no matching of devices (I abhor that; yes, I know, I'm not an engineer)
* decent psrr
* ??? what else ???

I really don't know what else would I want from a unity gain buffer? Please, experienced designers, what say you? What does one want, realistically, from a really good buffer?

That being said, let's start with something that I'm considering. The attachment shows a very simple buffer. A follower of sorts, a current source there too. I'm sure much better has been done, but it would be nice to have a simple circuit. Have not prototyped this yet, just simulated results.

First attachment, the schematic, which is no surprise; this has been done everywhere, it's not sophisticated.

Second attachment, Vcc PSRR.

Third attachment, Vee PSRR.

Fourth attachment, frequency response.

Fifth attachment, distortion number, 1V p-p input, into a 63R load, at 1kHz, 215mA idle current. With higher idle current the distortion becomes lower.

[Elvee]

You could opt for a Tringlinator:
http://www.diyaudio.com/forums/headp...amplifier.html
The distortion for the complete amplifier is slightly lower at 5x the level and output load halved (32ohm).
The distortion of the tringlotron buffer alone is even lower, and it only requires 4 active elements (6 if the input buffer is included).

[saurus]

I use this one in one of my headphone amps. simple, and it sounds very good:

Audio Buffer in english

[ikoflexer]

Originally Posted by Elvee

You could opt for a Tringlinator:
http://www.diyaudio.com/forums/headp...amplifier.html
The distortion for the complete amplifier is slightly lower at 5x the level and output load halved (32ohm).
The distortion of the tringlotron buffer alone is even lower, and it only requires 4 active elements (6 if the input buffer is included).

I will look into it, thanks! Do you think that the 3rd order distortion will be lower than the circuit I posted? Also, Elvee, how can you get rid of that nasty output capacitor. For my schematic I can easily implement a servo so there's no need for output capacitor.

@saurus: I would like to not use any opamps (except possibly in a servo).

[ikoflexer]

This is what I have in mind for the servo, just a typical loop and no large cap at the output.

[jcx]

what headphone, ohms and sensitivity - sensitivity varies orders of magnitude so its an important spec to have for the amp design

if you really don't want to select, trim then jfets really aren't the parts for you - standard grades can have 3:1 parameter spread

having the input jfet bias depend on BJT beta is likely not so good
a b-e resistor is at least approximately constant current

why introduce the mosfet? - another bjt same part # as the output Q would work fine
likewize cascoded jfet ccs bias for the output ccs bias seems a bit overdone if you're not at least cascoding the output Q too

[ikoflexer]

I got the Sony MDR-V6, not exactly audiophile quality, but fine for me. 106dB/mW sensitivity, 63 ohms at 1kHz impedance.

Didn't know an easier way to make it high input impedance and also get rid of high input capacitance.

The mosfet CCS seems to work better then a bjt, as a simple CCS. The jfet ccs can be replaced with a resistor, but the results seem better with the jfets.

[Elvee]

Originally Posted by ikoflexer

I will look into it, thanks! Do you think that the 3rd order distortion will be lower than the circuit I posted?

I don't think so: it will be slightly higher or similar.

Also, Elvee, how can you get rid of that nasty output capacitor. For my schematic I can easily implement a servo so there's no need for output capacitor.

I am afraid it is impossible without inordinate additional complexity.

[ikoflexer]

Thank you for your suggestions guys! I tried to change the circuit a bit, and I think I like the change. It's even a bit simpler. 2nd order harmonic distortion has increased a bit, but I don't think I care. 3rd order harmonic distortion has decreased a bit. This is only simulation. Hopefully reality will not be too far from this.

[ikoflexer]

Hm, there's this variation, which has a weird effect lowering the odd harmonic distortions. It looks interesting enough to prototype this circuit and see what it does in real life.