Cheap 24/96 DAC with CS43122

[jwb]

Since that other thread disintegrated, I'll post my stuff here. Here is a schematic and layout for a 24-bit, 96kHz, stereo DAC. You should be able to build this for less than $300. The board is run from a +/-10V supply fed in through the molex connector at the top. The 16-gauge output pads are, from top to bottom, R-, R+, L+, L-. Coupling capacitors and output resistors are needed, but I left them off so you can make them to your taste. Note also that I have neglected all inputs. Hook them up to pins 4 and 5 on the CS8420 as you wish, and place them in the vast empty space on the board's periphery.

Many features are left as an exercise for the reader. You could make this DAC better, and you could make it cheaper. Please post any modifications you make here!

NB: this design is untested, as I have not built it and in fact have no plans to build it. I drew it only because I was intrigued by peranders' idea of a DAC costing less than $300.

schematic
Gerber files

[hifi]

I could not open the gerber files...=/

[peranders]

Originally posted by jwb
Since that other thread disintegrated

I surely agree!

Looks nice, your pcb, but what about analog and digital groundplanes? Crystal claims that this is really important in order to get high S/N. The thing that worries me the most is in fact EMI. These nasty waves seems to wander through my LF-section despite the fact of having separate groundplanes and separate power supplies.

[peranders]

Your "minimalistic" solution can be a good start of a decent DAC.

What is a bad DAC?

What is a fair DAC and what is an excellent DAC?

Some thinks that only Burr-Brown fits but I have noticed that Crystal and AKM are very common in professional gear, so they can't be that bad. After all, many CD's are created with Crystal and AKM in the signal chain. Just a thought.

300 USD is very realistic if you are lucky, home made pcb's, cheap case, free samples, junkbox full of junk etc.

[hifi]

The outputstege is it tested in some other dac design or is it just conceptual?

/micke

[Jocko Homo]

I couldn't open the Gerbers either. Is there an aperature file that we are missing?

I've seen similar I/V schematics up here before. I suggest that you might want to test it first, as some have distortion numbers higher than you might want to accept.

Jocko

[hifi]

Well there is no need for I/V conversion with the Cs43122 since it outputs voltage. a buffer should suffice? maybe even transformer coupling?

The LT1962 seems quite hard to find what about somthing simpler like a TL431 ?

In my opinion the PCB should be made to be easily made at home that means no traces thinner than 1mm not to many vias. and only 2sides.

separate groundplanes should not be so hard to make.

Is the AKM dac voltage out? than maybe some form of adaptercard can be used for the dac to make it mor universal?

just my 2c

/micke

btw...is this still peranders show or is it a diyaudio.com project now?

[jwb]

Originally posted by Jocko Homo
I couldn't open the Gerbers either. Is there an aperature file that we are missing?

The file is for Gerber RS274X. The apertures are embedded. I can view them with gerbv (linux) and with Viewmate (windows). The drill file is missing. I'll add it when I get back to that computer.

I've seen similar I/V schematics up here before. I suggest that you might want to test it first, as some have distortion numbers higher than you might want to accept.[/B]

This isn't an I/V scheme, it's just a buffer and line driver. The output from the DAC is voltage, but I wouldn't want to drive a cable with it. In the world of source followers, I think this one is pretty common.

[jwb]

Originally posted by peranders
Looks nice, your pcb, but what about analog and digital groundplanes?

You can slot the groud plane around the output buffer and the topmost regulator, but you must leave it connected beneath the DAC. The problem will be that two +10V power traces will cross this slot, and their return path to ground will be directly beneath the DAC. That would be worse, I suspect.

The solution is probably to bring in +10/GND/-10 on the top connector, and add a connector on the bottom for +10/GND. Cut the power traces that cross the slot.

Crystal claims that this is really important in order to get high S/N. The thing that worries me the most is in fact EMI. These nasty waves seems to wander through my LF-section despite the fact of having separate groundplanes and separate power supplies.

I think the key here is not to separate the ground planes, but to consider all the return paths for every signal you route. In this layout, I made certain there are no high-speed digital signals with return paths through the analog section. In fact all the digital crap is confined to a small area in the center of the board. The +5V rails are all decoupled with capacitors practically atop the power pin, and I have provided a nice fat via to ground from each ground pin. The +10V rail feeds the regulators via ferrite beads (L1, L2, and L3), so I expect the +10V rail to be quiet.

As I said, this is a simple DAC with embellishments left to the reader. A split ground plane and independent supplies with separate transformers would be one of those embellishments.

[jwb]

Originally posted by hifi
In my opinion the PCB should be made to be easily made at home that means no traces thinner than 1mm not to many vias. and only 2sides.

You can't have these constraints and work with these parts. The pads on SOIC are .5mm wide. You can't route to them with 1mm traces. You'll notice I had to route traces beneath the SOICs. There isn't enough room for four 1mm traces in there.

As for vias, they are critical to performance. Every IC ground pin must have a low impedance path to the ground plane. Every decoupling capacitor, or group of capacitors, should have a via to ground as well. I think I kept vias to a minimum here. There are no vias on signal or power traces, only vias to ground, and I have used only .050", .040", and .020" drills.

btw...is this still peranders show or is it a diyaudio.com project now?

I told you. We're an anarcho-syndicalist commune. We take it in turns to act as a sort of executive officer for the week. But all the decisions of that officer have to be ratified at a special biweekly meeting. By a simple majority in the case of purely internal affairs, but by a two-thirds majority in the case of more ...