Modular multi channel DIY AD/DA Box

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I really didnt mean to offend anyone by my comments :( , im just going by what i heard from the samples posted. I gave an unbiased opinion on what ive heard. The lavry gold is one of the finest converters made, but what i heard was almost as good.
 
Ok, to make some progress, here is the schematic of the DAC module. Please note, that the values of the output filters are not correct and that I used the NE5532 but you can also use another opamp that comes with the same pinout. I think I'll use the AD8620.
I'm now going to order prototypes of the DAC pcbs as soon as I have found a good source for the PCM4104 in small quantities.

Raphael
 

Attachments

  • PCM4104_module.jpg
    PCM4104_module.jpg
    295.5 KB · Views: 1,079
rkn80 said:
I'm now going to order prototypes of the DAC pcbs as soon as I have found a good source for the PCM4104 in small quantities.

A big part of the performance will be in the layout; if you post it you could get some useful feedback.

Sources for the PCM4104 include Digi-Key and Mouser; please put your location in your profile for possible local suppliers.

JDB.
[and there's always TI's samples program]
 
The layout I'll post today evening when I'm back at home.
I tried the TI sample program but they do not sample the PCM4104.

Raphael
 
What a luck: I just remembered that I've the pcb project on my MacBook too, so here is the layout of one DAC module (4channels).

Raphael
 

Attachments

  • PCM4104.png
    PCM4104.png
    87.2 KB · Views: 842
rkn80 said:
here is the layout of one DAC module (4channels).

Looks quite reasonable. The only issues I have are relatively minor:

- the +5VA trace running under the chip makes a nasty moat in the ground plane. I would suggest either stitching the top-layer ground under the chip to the plane with a few vias, or even using a single jumper wire and keeping the ground plane unbroken (ugly but functional).
- I wonder what happens wrt ground loops given that you have grounds connected at both X101 and X1001
- consider removing some of the ground plane right below the inputs of the op-amps and their associated traces, this will help stability especially with faster op-amps.
- those capacitors marked X7R will be C0G/NP0 in the final version, right?
- the MCLK trace is relatively long and unshielded
- you may want to directly connect the ground on C1018 to the ground side of IC1002

Other than that only a test will show its performance. It looks interesting, but definitely not a soldering project for the average DIYer.

JDB.
 
jdbakker said:
- the +5VA trace running under the chip makes a nasty moat in the ground plane. I would suggest either stitching the top-layer ground under the chip to the plane with a few vias, or even using a single jumper wire and keeping the ground plane unbroken (ugly but functional).
Yes, routing the +5VA was/is a problem on a two-layer design. First I routed it under the balanced output lines, but then the analogue outputs a crossing a split in the ground plane. So I decided to place a copper plane under the chip on the top layer und to run the 5VA tracks under the chip on the bottom layer.

jdbakker said:
- I wonder what happens wrt ground loops given that you have grounds connected at both X101 and X1001
The reason that you have gnd lines on both connectors is that you then can connect flatband ribbon cables without using a mainboard, you can go directly to a sub-d or XLR connectors. Indeed if you want to plug in this module in a mainboard you have to make sure that you are not making ground loops. So think of it like offering an additional option.

jdbakker said:
- consider removing some of the ground plane right below the inputs of the op-amps and their associated traces, this will help stability especially with faster op-amps.
Never seen that on a solid ground plane.

jdbakker said:
- those capacitors marked X7R will be C0G/NP0 in the final version, right?
Yes. As I said the values in the filter are not correct. I simply used what I found in my component library. :)

jdbakker said:
It looks interesting, but definitely not a soldering project for the average DIYer.

Hey, come on. Its only 0805 and TQFP. My first soldering projects have been already made with these case types. It is not so difficult like it seems to be. You need to know two or three tricks. :)

Raphael
 
rkn80 said:
jdbakker said:
- consider removing some of the ground plane right below the inputs of the op-amps and their associated traces, this will help stability especially with faster op-amps.
Never seen that on a solid ground plane.

For some of the faster op-amps that some people like to use for audio work nowadays it is pretty much required. Have a look at the TPA6120A2 data sheet starting at page 16 for a brief explanation. (I know that the '6120 won't fit on your board, but the theory remains the same, and applies to both VFB and CFB amps).

For 5532-class op-amps this is generally not necessary (depending on topology, layout and impedances, of course).

rkn80 said:
jdbakker said:
It looks interesting, but definitely not a soldering project for the average DIYer.

Hey, come on. Its only 0805 and TQFP. My first soldering projects have been already made with these case types. It is not so difficult like it seems to be. You need to know two or three tricks. :)

I know that you can do quite a bit with minimal tools and not too much effort, but recent discussions show that many people are still quite cautious toward even simple SMD work. Besides that, your board doesn't make life extremely easy: the pads of the passives don't look to be too large, in some areas the passives are packed rather tightly (good for performance, bad for novices). I'm not new to SMD work, but I still hate hand-soldering D-PAK/TO-252 parts like your voltage regulator.

JDB.
[could be worse, could be a TO-263 on a nice big thermal plane. Brrrr....]
 
Why so pessimistic? Soldering the passive parts is not more diffcult then wired parts. It is even fast because you do not need to prepare the wires etc... The only thing, that is a littlebit difficult is the DAC but there is no way around because you can only buy it in this package.
I do all my DIY stuff in SMD and I like it. If you etch the PCB by yourself you save a lot of drilling work and you win a more solid ground plane.
I think, I'll make an instruction manual when I'm building the prototype to support people who are scared by SMD. ;) I've no special tools for it. I only use a solder iron with a small tip and that works quite fine.
I agree it becomes very hard or impossible when we want to use BGA or packages without exposed pins. But these types I normally avoid because I know that I cannot solder them myself.  ::)

Raphael
 
rkn80 said:
I agree it becomes very hard or impossible when we want to use BGA or packages without exposed pins. But these types I normally avoid because I know that I cannot solder them myself.  ::)

Sure you can. Just use solder paste, a stencil and an oven or a hot plate.

JDB.
[and people have etched their own stencils, too, although considering the cost of the parts I prefer buying laser-cut stencils]
 
if this project really gets going, I may be of some help... my father owns a circuit board manufacturing company (which I have worked in), with solder paste screening, pick and place machines, and a couple of reflow ovens.
 
iangomes said:
if this project really gets going, I may be of some help... my father owns a circuit board manufacturing company (which I have worked in), with solder paste screening, pick and place machines, and a couple of reflow ovens.
This migth be cool, if it is not too expensive. Anyway, soldering is not the most fun part of dying for me, I might be happy to purchase some prepared boards.

Michael
 
I made some minor changes on the layout (gnd vias now without thermal relief etc.) and ordered the first prototype board of the DAC today. It should arrive in 8days.

I'm think of making a ADAT input board first before releasing the ADC  board because then there is already use for some people: You can put the board together to make a nice multi-channel DA converter. What do you think?

Raphael
 
This project looks very nice !
I'm definitevely interested in DA with adat with nice ouput opamps, cheap and reliable.
What would be nice also is to balance via That Chip to the outside world (with a nice output TX behind it, it's just creamier !)

Cheers
 
Yeah, Adat is perfect.

I got a RME adat sound card with 3X 8 light pipe input/output, I've been waiting for ages for a DIY converter project.  I will definitely need some of these boards...

Thanks Raphael.

J
 
Back
Top