DIY USB-1543 Dac Kit Help

[recca]

Hi all. I'm a total newb to DIY but have just ordered by first project. Its a DIY USB-1543 DAC Kit, the AW-D# on this site: http://www.audioworkshop.com.hk/index_en.htm

I've been searching the site for possibe upgrades to this before I put it together. Here are some thoughts I've had:

1. External power supply for the DAC Section. Battery use the regulator section built into the board? I get the impression that battery will be best, but I'm afraid that it will be harder for me to implement correctly.

2. DAC Paralleling. Seems like 8x1543 will result in excellent performance. I think it should work although I'm worried about the amount of power that will be needed for the tower. I guess adding a better power supply for the DAC section will be necessary if I go that route.

3. I/V conversion. I honestly have no clue on how this is setup on this board. I get the impression that passive IV works best. Is this a good implementation? Any suggestions on how to improve. I'll like use hookup wire to connect to RCA's rather than use the included RCA's.

Thanks in advance for the help!

 

[John Swenson]

Aha, these kits are in now.

I've built several 2706 based DACs from scratch, one of which uses an 8X stacked 1543. It sounds quite good (better than anything else I'd ever had). You can do better although that takes a LOT more effort.

I would recommend building it stock (external RCAs are fine) and letting things burn in and get familiar with the stock sound before making changes. Definately build it with the esternal supply powering the DAC.

The changes I would make are:

running the 2706 off the external supply, getting it off the USB bus is a big improvement. Its a very simple change to make so I would keep it stock, then make the change later.

Try different output coupling caps, the 100uf electrolytics are not really the best way to go, I personally would either use something like a viaminQ or a good polypropelyne. If you use the external RCAs its real easy to swap caps.

Try different types and brands of resistors on the outputs of the 1543, this can make a huge difference. I was totally amazed at the sound change by going to different types of resistors. This is one of the beauties of DIY, you can choose what sounds best to YOU in your system.

Use a Kwak Clock on the clock input instead of the crystal. This takes an extra board and is not trivial to implement, but it does significntly improve the sound. (I wish Guido made a 12.00 MHz clock)

Go with the 8X parallel DACs. This takes 1/2 amp and would probably stress the power supply on the board. Making the DAC tower is not trivial either.

BTW I see you are in San Francisco, I'm in Fremont, I would be glad to help with your project and I would love to hear it when its done. (If you like you can come over and listen to mine).

John S.

 

[recca]

John

Thanks for the reply! They are in stock and I was lucky to have one sent. They just sent it today so I think I should get it next week.

I like your suggestions. I suppose building as it should be will be a great learning tool for me. I got excitied looking over everyone's mods and was hoping to have mine tricked out too.

I'd love to build the 8 DAC tower, but I don't think I'm ready for that, plus i have no clue what kind of case that would fit in!

I shot an email out to you but don't know if you received it.

peace. . .

 

[Guido Tent]

(I wish Guido made a 12.00 MHz clock)

John S.

Hi

www.Dddac.de uses 11.2896 which works fine

In addition, I also have 12.288 MHz, if that helps

 

[Brent Welke]

12 MHz Clock

I use a 12 MHz clock from Mouser. I don't know how that compares to a discrete clock circuit but it sounds better then using a crystal.

See this thread for details:
http://www.diyhifi.org/forums/viewtopic.php?p=6569#6569

 

[John Swenson]

Hello Guido, Doede is using your 11.2896 clock with a 8412 which has a mode that allows an external clock to be use to clock data out. We are talking about a 2706 USB interface chip which takes a 12.000 MHz clock. The spec sheet says it needs 12.000MHz +- 100ppm, I think 11.2896 is a little out of that spec.

It does not have any means of taking an external clock in to clock out the data. If you feed it 44.1 you DO get a 11.2896 clock out on its sysclk pin but its not synchronized to any external clock.

So far we have been talking about using a low jitter 12.00MHz clock to feed the 2706, it DOES improve the sonics considerably without resorting to ASRCs etc. A traditional FIFO approach using your 11.2896 clock would probably work, the 11.2896 coming out of the 2706 seems to be much more stable than what you see out of most transports so the drift between it and the external clock probably will not be too great making the FIFO circuit fairly easy.

But so far I've just been trying to keep this simple, if giving the 2706 a low jitter clock significantly improves the sound then that is what I'm going to go for first before trying much more complicated circuits.

BTW I've tried MANY different DAC chips and digital filter/NOS combinations recently. These all produce significantly different sounds. So far my favorite has been PCM1704 driven by a NOS circuit that alligns the channels (the simple non-alligned version did not sound nearly as good).

I've tried resynching the signals sent to the DAC using the sysclk from the 2706 and this significantly improves the sound. This leads me to believe that sysclk coming out of the 2706 really is a pretty good clk.

John S.

 

[Guido Tent]

Hello Guido, Doede is using your 11.2896 clock with a 8412 which has a mode that allows an external clock to be use to clock data out. We are talking about a 2706 USB interface chip which takes a 12.000 MHz clock. The spec sheet says it needs 12.000MHz +- 100ppm, I think 11.2896 is a little out of that spec.

It does not have any means of taking an external clock in to clock out the data. If you feed it 44.1 you DO get a 11.2896 clock out on its sysclk pin but its not synchronized to any external clock.

So far we have been talking about using a low jitter 12.00MHz clock to feed the 2706, it DOES improve the sonics considerably without resorting to ASRCs etc. A traditional FIFO approach using your 11.2896 clock would probably work, the 11.2896 coming out of the 2706 seems to be much more stable than what you see out of most transports so the drift between it and the external clock probably will not be too great making the FIFO circuit fairly easy.

But so far I've just been trying to keep this simple, if giving the 2706 a low jitter clock significantly improves the sound then that is what I'm going to go for first before trying much more complicated circuits.

BTW I've tried MANY different DAC chips and digital filter/NOS combinations recently. These all produce significantly different sounds. So far my favorite has been PCM1704 driven by a NOS circuit that alligns the channels (the simple non-alligned version did not sound nearly as good).

I've tried resynching the signals sent to the DAC using the sysclk from the 2706 and this significantly improves the sound. This leads me to believe that sysclk coming out of the 2706 really is a pretty good clk.

John S.

Hello John

I agree with above. After all, the DAC clock quality is of highest importance, which probably explains why reclocking the PM1704 signals helps.

I'll consider taking the 12MHz clock into production.

best

 

[Nisbeth]

I'll consider taking the 12MHz clock into production.

Would be very nice as all TI's USB-chips use a 12 MHz clock


/U.