My latest project - DAC + headphone amp

[gmarsh]

I fixed the last software bug on this today, so I'm officially calling this project "done". The goal was to create the highest quality, most transparent headphone amplifier possible with little comprimise.

- The signal path is TORX142L or AD8611 -> CS8416 -> AD1896 -> PCM1792 -> OPA4134 -> TPA6120A2.

- There's also an ATMega8L onboard which performs digital volume control - the volume knob feeds the Mega8's ADC a voltage, which digitizes/filters the sample and updates the digital attenuation register in the PCM1792. The Mega8 also handles automatic input selection, swapping back and forth between optical/coaxial inputs until the CS8416 locks.

- Power supplies (3.3V digital, 5V analog, +-12V analog) are regulated using LM1085 regulators. Rails and grounds have solid planes. Analog/digital grounds are connected using a 0-ohm resistor underneath the DAC.

- The oscillator is a cheap 5x7mm SMT XO from CTS. It measures better than any sort of XO/VCXO i've ever constructed.

- I'm using the AD8611 SPDIF circuit I made a thread about a while ago, which uses hysteresis (25mV) for stability whenever there's no input connected. Sorry Jocko

- Other parts... Case is a Hammond 1455L1601BK, front/rear panels are made by frontpanelexpress.com, headphone jack is a slimline Neutrik with gold contacts, and the spun aluminum knob is from Kilo.

How does it sound? Can't say anything bad about it. I love how I can have the volume cranked up all the way in a completely silent room, and I can't hear anything with no input playing. And with music playing, the sound is crystal clear... right now I'm listening to Mule Variations, and I don't know if it's placebo effect but I'm noticing all sorts of different instruments and background things that I never noticed before. Tom Waits makes a great sonic treasure hunt.

I'll test it on the Audio Precision at work sometime, which has better ears than I do.

On with the pictures:

front view:


rear view:


4-layer, bare PCB:


The assembled PCB, during software hacking:


And finally, here's the entire set of build files (gerbers, eagle files/libraries, etc) - feel free to tear them apart and critique me, or rip them off and build your own. Just don't make a million dollars off the design without buying me a beer.

http://rapidshare.de/files/2470563/HD1.zip.html

[300_baud]

are u selling PCBs?

[gmarsh]

Originally posted by 300_baud
are u selling PCBs?

hadn't planned on it, but I do have a spare PCB i might sell if anyone wants it... $50 USD? (it cost me $122 for the two originally)

[rlim]

Where did you have those boards fabricated? Sounds like a great price for a 4-sided board.

[gmarsh]

Originally posted by rlim
Where did you have those boards fabricated? Sounds like a great price for a 4-sided board.

Sierra Proto Express... "No Touch" service. $51 each (min 2) + $20 setup.

[tobias_svensk]

dude, that is probably one of the nicest works i've seen!

[tobias_svensk]

Is that spare board still at sale?

[Robzy]

Holly ****! Im mighty impressed! Just a couple of questions:

How hard was it to program the microchip?

and

Why do you need the resampler? Couldnt you just pass the output of the CS6416 onto the TI DAC?

Rob.

[gmarsh]

Originally posted by tobias_svensk
Is that spare board still at sale?

nobody's asked me for it yet... want it?

Originally posted by Robzy
Holly ****! Im mighty impressed! Just a couple of questions:

How hard was it to program the microchip?

and

Why do you need the resampler? Couldnt you just pass the output of the CS6416 onto the TI DAC?

Rob.

Programming wasn't hard. I soldered the unprogrammed Mega8L directly to the board and programmed it via a 6-pin header with an AVR ISP programmer... If you don't have an ISP, you can build a Ponyprog.

I'm using the AD1986 for jitter rejection - its sample rate tracking loop is far superior to the PLL loop characteristics of the CS8416.

[Robzy]

Originally posted by gmarsh

I'm using the AD1986 for jitter rejection - its sample rate tracking loop is far superior to the PLL loop characteristics of the CS8416.

I was actually refering to the SRC4192, and i cant seem to find the AD1986 on your board anyway... :S.

Still, im flabbergasted at the design.

Rob.