AMB Laboratories DIY Audio

[linux-works]

ok, I'm not really serious about the name for it

the mounting is obviously not serious but it sure did let me eval this chip. the underlying platform is the 'usb audio widget' which converts usb to 24/192k via 'audio class 2' usb (no longer limited at 44.1/16). I'm tapping into the i2s lines, 5v power, analog-out and gnd. the ribbon cable carries i2s from the big SoC chip (atmel thing) and the thin blue wire is the system clock (not bitclock).

lm1117 3.3v reg chips are used (3 of them!). lots and lots of bypasses as per the spec sheet.

what's cool is that this is a dc-coupled direct drive output, no caps, no filter, no nothing. no bipolar supply needed as the negative rail is via an onboard charge pump. 2v level on output, too.

it took most of the day, today, to solder this up. I just heard audio come out about half an hour ago - what a feeling





the DAC chip is a few dollars and specs are on mouser/digikey/TI sites.

not a high end DAC but a very simple to implement DAC!

[Shaman]

Nice work, LW!
Also check ESS 9022/9023 when you have time. You'll be surprised.

[linux-works]

that dac board uses that ESS chip. its a super nice chip, too! many people consider them similar sounding and they are similar priced, too.

the ESS is harder to work with since it NEEDS 2 local clocks. the ti chip, otoh, not only does not need all those clocks, it derives the system clock from the datastream automatically! one less high speed data line you have to run, one less thing to bleed thru and create noise or jitter. the ti chip is buyable in single units whereas the ess has to be bought from one place (only) and in 25 units each time. I do like the ess lead spacing for soldering (lo) and wish ti had gone a bit lower density on the pins.

I would be happy, as a user, with either chip. but to build or stock or sell, the TI gets my vote.

[Shaman]

Truth be told, it mustn't get harder to implement than the Sabres...
I couldn't appreciate it's qualities until I got a carefully optimized ready-built unit from EUVL.
The exact opposite of the disaster called NuForce uDAC....

[linux-works]

I have not followed the 'story' on nuforce. what's the tl;dr version?

[linux-works]

I just tested 'self clocking' mode and that works, great, too.

you can disconnect that thin blue wire that is going to the xtal clocks on the main board. the ESS dac needs that but the TI dac can 'figure out' what the right system clock should be for the data rate. it sets itself so you don't have to (lol). that's one less wire you have to run (even more important if you run this between boxes!) and one less clock source to have to worry about.

to set it to self-clock you remove that blue wire and connect it to gnd. now, all the i2s data comes over that ribbon of green/yellow/orange.

[linux-works]

new variation:



using a fiio D3 as the i2s source (it receives spdif via opto or coax, into a wm8805 chip). tapped into it via my ribbon cable (extremely hard to do that; took me many tries!), tapped into its usb 5v and that was that. music came out!

I'll have this with me at the headfi meet, in about a week from now.

[ds23man]

Cool!

Regards Gerhard

[Shaman]

I have not followed the 'story' on nuforce. what's the tl;dr version?

Oh, I was just disappointed with it's performance.
After I had sold it, that "Objective" guy showed some measurements which explained the disaster in detail.
You'll find it in his blog.

[linux-works]

I can only tolerate a few minutes of reading nwavguy...

"why say something in 100 words when you can bore people to death with 10,000?"

sigh.

...I just get lost in his verbosity he really needs to trim his rantings down and de-emotionalize things a bit more. oh well.

also surprising that there were no internal photos of that dac. lots of photos of scope traces but I could not find an internal photo.

I guess the main complaints were distortion when the signal hit 0db and also analog pot imbalance (?)