DIYHiFi.org

Hallo Klaus,

I hope you still check this thread!

I`m about to get the parts I need for my USB DAC (chip is already here). I am also in germany, perhaps you could give me some help with the following:

1) What schematic did you use (I think J. Swenson pointend at the schematic on the 2702 spec sheet)?
2) Where did you get your crystal resonator from? Are there big quality differences here?
3) How long can the 2706 and the 1543 work with the small alkalines? Altough supposedly noisier I`ll be using rechargables.
4) are you paralleling the caps as suggested in the 2702 spec sheet
5) I think you used the be453 adapter fron conrad.de, how difficult was soldering?
6) What res. are you using for I/V (make and value)?

Thanks for your suggestions,

Sebastian

John Swenson suggested I try this clock, I did and wow.
http://www.mouser.com/index.cfm?handler ... deid=54925

It takes it to a new level of performance. It is simple to use as it also runs at 3.3 volts. To use it remove everything from pins 12 and 13. Attach the clock output to pin 12 and also a 33pF cap to ground. Pin 13 has nothing connected to it.

With self power (not USB powered), this clock, TDA1543, passive I/V, Opa627 buffer (sorry Jocko, I haven't tried your method yet) and gainclone is a killer combo.

On Aug27th I will take my prototype to a local diy audio event for some feedback on it. If it is well received I will make a kit/board for it to help spread the goodness :)

~~~~~

Ochoa: be453 adapter fron conrad.de works geat. Soldering is easy, have some desoldering wick handy, you will need it. I/V = 2.7K Riken .5 watt and 4.7 Black Gate N, or Black Gate 22uF Hi-Q are ok too. Schematic to use is at the beginning of this tread. Only change is to join pins 10 and 11.

here is a kit in the works. Similar to to what I would design.
http://www.audioworkshop.com.hk/report/d3/d3.html

On the home page they say available mid Aug.

Hello Sebastian,

All the parts can be bought at Conrad except the TDA1543 which I got from Reichelt.

1) What schematic did you use (I think J. Swenson pointend at the schematic on the 2702 spec sheet)?

I used the schematic from the data sheet on page 29 of the PCM2707. Here is the version in an easier to red form.


2) Where did you get your crystal resonator from? Are there big quality differences here?

I have no clue if there are any differences in resonators, I just bought mine at Conrad, but I will be using what Brent uses.

3) How long can the 2706 and the 1543 work with the small alkalines? Altough supposedly noisier I`ll be using rechargables.

The alkalines were only a test setup. I tapped in the 3.3 V from the coputers SMPS and that worked pretty good. With some filtering that might be the ideal solution because any computer has a 3.3V line.

4) are you paralleling the caps as suggested in the 2702 spec sheet

Yes, but as in spec sheet of 2707

5) I think you used the be453 adapter fron conrad.de, how difficult was soldering?

I was concerned about soldering, but it is very easy. Just solder all the pins on one side at ones not paying attention if the solder flows between the feet. Do all the four sides like that and then use small solder wick with some soldering flux on it and pull of the excess solder. Then use a small cutter nife and scratch between the pins to make shure no solder remained.

6) What res. are you using for I/V (make and value)?

Peter Daniels
As a future project:
http://www.dhtrob.com/projecten/nonodac_en.htm

There is some issue with the PCM2706/07
Since I did an update on my XP the computer dosn`t recognize the USB device anymore. It worked flawlessly up to that poin. I have build and tried several devices because I thought at first that I killed the chip some how, but I had no luck. So I assume that it is a compatibility problem with at least the German version of XP. It is stated on the data sheed that the PCM2707 has been changed in May this year to make it compatible to XP Global, before that time it was only for the English and Japanese XP versions. Either the update did not work for the German XP version, or the samples I got last month are still pre May 05 versions.

Can you give me your phone number?

Klaus

Brentw:
Thanks for your help. I`ll look forward to your succes in the diy show! By the way, would you be designing the integrated gaiclone with 1543 and 2706? That would be a really nice poyect!

Hi Klaus, thanks for the information. One thing, are you ordering that clock from mouser or are you sourcing it in Germany? Have you already ordered it?

Could you send me a PM? I?m in T?bingen, but on thursday all be leaving for a few weeks and after coming I?ll be moving so I have no tel nuber to give you at this time. You can contact me through my email anytime.

Gr?sse
Sebastian

I finished the prototype for the audio event tonight. Now it has till Saturday to burn-in.

I snapped 3 pics of it.

pic 2

last pic.

The bottom board is a Peter Daniel board. It made a good starting point for this prototype. On top is two boards. The PCM2706 on the adapter board and next to it a OPA627 buffer and volume control. The volume control is before the buffer.

The black cable is the USB cable.
The blue and orange wire from the top board is the output after the buffer.
The blue and orange wire from the bottom board is an output before the buffer and volume control. I don't use this output but it may come in handy at the event on Saturday.

I ran out of BG caps for the buffer board, that is why you see a couple Panasonics on it. I hope more caps will arrive before Saturday. One channel uses the BG and one the Panasonics. I don't hear a difference yet.

It sounds pretty nice, much less noise then my first prototype and one less transformer. I think it will get a good review at the event.

~~~~~~~~~~

I use the buffered output with volume control and feed that to a couple gain clones. I don't think I would add a gain clone to this DAC board. It is simple enough to build them seperately and just connect them, more flexible too when trying different amps. Who can resist mix and matching gear :)

I am thinking about making a 8 channel version with an on board USB hub, but that hasn't been worked out yet. I am still working towards a digital XO. One step at a time I will get there.

Brent, what was the response to your USB Dac at your recent audio event?

Thanks for asking.

The event ... http://www.diyaudio.com/forums/showthre ... genumber=5

Not so good. I didn't have it running very good at the time. The night before I was still soldering in Black Gate caps. Also I knew there was some digital noise but didn't have time to solve that. It was a mess. The next day I noticed one of the 4 regulators for the OPA627 buffer was shorted making one channel louder then the other and distorting. I didn't even try the unbuffered output at the event, I should have but I was a bit nervous about showing it off after the buffered output didn't sound so good. In a couple weeks Jason (new DAC in the mail), Chris (Jolida CD player) and I are getting together to have a CD player/DAC shoot-out. I will report back as to how that goes.

At the event there were 20+ pairs of speakers to test out. It made sense to find the best source in at the event and stick with it. A Jolida 603 CD player was the best source to many (me included). The Jolida has a tube output. I don't know if that is what makes it sound so 3D but I think it is time to give a tube buffer a try.

... And the jitter was among the main things I wanted to know about so I've ran the test using the -3dB 11025Hz signal toggled by the 229.6875Hz LSB. It took some time to isolate the data related sidebands (marked in red). And these are looking much better than I've been prepared to see. In fact they are solely related to the TDA1543 which showed known string of the data related sidebands increasing all the way down to 229.6875Hz with levels that match rather the sub 16-bit resolution. Those not aware of the jitter performance of the TDA1543 can find it here and here. In fact, the first sidebands (+/-229Hz) are now about 8dB lower than when I've been driving the TDA1543 by Marantz CD-63?s S/PDIF output and via CS8412. So the USB doesn't appear here as a bottleneck.

Looking across the wider bandwidth the things look this way. The right string of data induced sidebands is generally a few dB calmer than the left shown here (the right string is also somewhat buried in the noise floor of my measurement set up since its noise shaping looses its steam at higher frequencies).

The bad news are those 1kHz related artifacts. As I could figure up to now they are caused by the fact the USB sends data in packets every 1ms. The first thought about the artifacts at +/-100Hz (marked in blue) on the previous graph is the ground loop but explanation doesn't necessarily stops here. Sidebands marked in grey (I thought firstly they are at +/-125/62.5Hz but if I am not mistaken they are in fact +/-128/64Hz) are for me still a bit mysterious. As is the one at 7875Hz. And as are a few others minor ones that I'd ignore for now.

Though some problems obviously may be diagnosed, or at least in this implementation they may, I was, as I said, prepared to see worse jitter figure. Since this readiness originated from some previously published measurements, I conclude that such poor performance is mostly caused by the implementations and not by the USB interface as such.

It is interesting... The harmonic distortion of this DAC at 1kHz is 0.046% which is a bit better than what its older S/PDIF brother has achieved in the same 6V/1k circumstances (0.079%).

(Please ignore those 50Hz related spikes, I did used long cables, several meters both the USB and analog ones.)

However, taking into account earlier mentioned problem with 1kHz I've had some doubt about the distortion measurement done with the signal of the same frequency. So I moved the frequency a bit to confirm it or remove it. I've chosen randomly 836Hz. So yes, the measurement at 1kHz was influenced by the USB's data sending frequency but yet not in a fundamental way. The THD is 0.05%.