Yet Another Crystal DAC with Lundahl Outputs

[stokessd]

I've been working on a new DAC to replace my aging PCM63 based unit (I want one that doesn't have a tube stage that I can leave on all the time).

I have been working on one based on the CS8416 receiver, a CS8421 sample rate converter, and a CS4398 DAC chip. It uses a pair of Lundahl LL1690 1:1 transformers for the analog stage.

The chip compliment itself is somewhat like the $125 Gigawork DAC, although I'm using a LOT of surface mount parts. I've got a whole lot more power supplies for the various voltages needed and lots of capacitance. However the biggest difference is the transformer output stage.

The reason I'm posting here is that I'm going to have a few circuit boards professionally made and I'd be willing to get more made if others wanted one as well. I'm planning on making either a 3 layer board (with ground plane inside) or possibly a 4 layer board with a divided power plane as well.

The design is mostly surface mount parts, although I'm using fairly large surface mount components (the chips are the smallest parts on the board). So this is not for the soldering beginner.

Here's the schematic:

An externally hosted image should be here but it was not working when we last tested it.

Here's the incomplete layout (hey, I'm still working on it):
http://quadesl.com/diyaudio/Composite Drawing.pdf


Here's some features of my design:

4"x7" board
board-mount power transformer
HEXFRED rectifiers
Input pulse transformer for coax input (via BNC)
TosLink input (for my airport express)
Input selector jumper (route to front panel switch)
Sample rate converter bypass switch
DAC sample rate jumpers (for use with sample rate converter bypass)
Pair of Lundahl output transformers
Transformer float jumpers.



So does anybody want a blank board?

I'm getting a minimum of 5 made, but probably will get more like 25-50 of them made.

Please email me at: stokes@quadesl.com

Sheldon

 

[stokessd]

Damn, nothing but crickets with this thread...Does that mean that my design sucks and nobody has any interest?

Not to be deterred, I went off this weekend and built it. I've never etched a board with traces as fine as needed for TSSOPs. But somehow I managed it.

I have all the details and lots of pictures here if anybody is interested:
Sheldon’s World Blog Archive Another Audio Digital-to-Analod (DAC) is Born

Sheldon

 

[Bill Fuss]

That's a damned fine job. The only thing that stops me from asking for one is my ability to solder the chips in. I love the idea of simplifying the design for trafo implementation. I would go one step further for I2S input and eliminate the 8416, but that can be done easily.

Have you looked into improving the clock PS with possibly something along the lines of a finesse circuit (Wenzel Associates).

 

[Pano]

Very nice. I've used transformer output with a number of DAC chips and have always like the CS chips. The seems to take well to the transformer stage and have a strong sounding output.

You should really enjoy it!

 

[stokessd]

That's a damned fine job. The only thing that stops me from asking for one is my ability to solder the chips in. I love the idea of simplifying the design for trafo implementation. I would go one step further for I2S input and eliminate the 8416, but that can be done easily.

Actually the chips were a whole lot easier than all the little 0604 caps and 0806 resistors. I use a totally sleezy but also very effective method to solder the chips in place. I've never heard it referenced before. Here's my secret:

I take a piece of scotch tape and attach it to the top of the chip extending over only one side of the pins (I know, ESD nightmare). I hold to the piece of tape to align the chip (it wants to settle between traces rather than on top of them). I then stick the tape to the PCB which holds the chip in place with only one side exposed for soldering. I hit the leads as best I can (I did these chips with an iron tip that was 2-3 leads wide), and don't worry about bridging leads, but making sure that each lead is soldered to the trace. I then pull the tape off (while holding the chip top applying pressure downwards). I then repeat along the other side of the chip. Now it's time to clean up my mess. I take some fine desoldering braid and suck up nearly all the solder. I then take a loupe and inspect. Works every time.

Hardly aerospace certified though.

Have you looked into improving the clock PS with possibly something along the lines of a finesse circuit (Wenzel Associates).

I have considered it, but I was reasonably happy with my clock output on the scope with the 0.1 uF cap next to the clock and the big honkin' tantalum next to that. But you are right, and that point is super critical. I was also thinking of dumping the film caps in the analog section (they are huge, and thus far away from the pins), and replacing them with ceramics, and doing a shunt regulator there as well.

I'm still scratching my head on the 24 bit sample rate conversion oddness.

Sheldon

 

[Pano]

If you have a hot air station, then this is not hard. Just buy the paste solder from Digikey or similar. Buy the smallest amount, it goes a long way!

Most chips and parts will pull themselves right into place. Fine pitch chips may bridge the pins. Use a thin solder wick to fix it.

 

[stokessd]

I should invest in a hot air station if I'm going to keep doing these sorts of things. Does the paste and hot air work well even if I don't have a solder mask? I make a lot of my own boards and I don't have solder masks.

Sheldon