So, I recently decided to design and build a headphone amplifier. I had long thought that an SS3602 discrete op amp driving a diamond buffer output stage would make for some killer head gear. After scribbling out some circuits and running them through a gambit of SPICE simulations, I found myself jamming components into a bread board.
I started out simulating and building the diamond buffer output stage. The Diamond Buffer is a well known output stage that lends itself to be driven directly from an op amp.
I chose to go with a CFP (Complimentary Feedback Pair) output device configuration over the EF (Emitter Follower) topology due to its superior linearity.
Its biased class A (naturally) with about 100mA idle current. The actual devices used in a final design will likely be different than what is used in this prototype. I built all of this with things I already had lying around the lab. For now the TO92’s are 2N5551 / 2N5401 and the TO220’s are MJE15032/33.
Once I got the output stage up and running and biased correctly, I then put the heart of the Amplifier, The SS3602 Discrete Op Amp onto the bread board.
The Discrete Op Amp is configured in non inverting mode with a gain of 4. At the end of the day, the gain will likely be increased, but for now I am using this with my Denon DCD1500 CD player (an 80’s vintage classic!) which has a 3V line output. So the gain of 4 is plenty in my current application.
I am a huge fan of DC servos for controlling offset as opposed to capacitive methods. So I saddled up a lowly TL072 onto a serf board for the task. TL072’s have inferior noise performance – certainly when compared to the SS3602 Discrete op Amp. So the output of the servo is ran through a 2 pole filter (the array of electrolytics) to clean it up before injection it back into theSS3602. I’m not a fan of electrolytics either, but since they aren’t directly in the audio path, and since I already had them on hand, I decided to use them anyway.
I would love to have some sort of film capacitors for this, but right now I just don’t have them around. The integration capacitors used in the servo are film type, since electrolyitcs leak too much.
The input section is definitely lacking. I just threw in a POT for volume control and ran it right into the SS3602 input. Ideally, this would be something like an Alps velvet potentiometer but that will just have to wait for my next Digikey order. If I wasn’t limited by bread board space I would have put a two pole filter on the input to set a bandwidth and filter out any RF that may get picked up in the input cables.
Ive also thought about taking the feedback from the SS3602’s output as opposed to having global feedback that encloses the diamond stage in the loop. This would give a higher output impedance and lower damping factor – and might even sound more tube like. Ive also thought about adding a bass and treble control – with a defeat switch, of course – but I like to goose the bass and treble slightly on some recordings. Something like a Baxadall tone control circuit may be in order. The problem is that its an inverting circuit by nature, and would require an additional stage to re – invert the signal back to being right side up.
Right now I’m running the amplifier from a bench supply. A finished design would run from a wall wort. If only they made split supply wall worts! I’m not really a fan of the “artificial ground” that is driven from some op amp. This requires the wall wort to be fully isolated. Even after all of that is hashed out, there are still the age old questions of “how many inputs?” and “should there be a line out?”
All in all, for a first ever prototype, the sound is quite good. It may be time to put together that Digikey order, and with any luck, this thing may just find its way onto the products page of my website.