[dsavitsk]

I am working on designing a non oversampling usb dac and could use some help with layout sugestions.

Here is the first go at a PCB.

Essentially, it is USB to PCM2707 to TDA1543. The I/V is just passive, and is a copy of the one used on diyparadise TDA1543 kit. There is also a Baby STEPS (TM) PS built in. The PS uses most of the steps design, but uses a LM7809 rather than an adjustable voltage regulator and also only uses one of the pre-rectifier filter caps and a very small version of the choke.

For PS connections to the pcm2707 I added space for 0805 decoupling caps figuring that I'd use 0.1uF ceramics, but I don't know if that's a good choice or not.

The 2 TO-92 things on the PCB are votage regulators (8V for the DAC and 3.3V for the PCM.)

Things that I know to be problems are dgnd is simply conected to the other grounds, and the connections to the PCM2707 got a bit messy.

The board is about 3.5" x 2.5" so there is not a lot more room. Any suggestions are eagerly anticipated, and I'll put up a schematic as soon as I draw one. For now, the only thing that is not self explanatory is the pinout of the pcm2707, but it is on the datasheet. You can also see this post from 00940 which is what I did.

Edit: fixed a couple of quick bugs in the layout.

 

[DaKi][er]

Ground plane, no reason to not use one

also, I wouldn’t have the transformer board mounted either, im a big fan of having the psu circuitry on the board but everything before the bridge rectifier would go, to have the transformer mounted as far as possible in the case from the board even if it is another 50mm or so. You shrink the board size to nearly 1/3 and also clear room from the analog circuitry from the transformer as well as giving much more option for power input - wall wart, toroidal or E/I transformer and keep mains voltage away from PCB's

After that it looks ok, I’m no fan of non-os or resistor i/v in either sound quality and in theory behind it but lots of people do like it

 

[robzy]

Quote:

Originally Posted by DaKi][er I wouldn’t have the transformer board mounted either, im a big fan of having the psu circuitry on the board but everything before the bridge rectifier would go

As above.

Rob.

 

[halman]

You have nowhere near enough clearance between the fuse and the chassies mounting holes.

There are two windings on that toroidal I recomend you dedicate one of them to the Tda1543.

And do use a groundplane but remeber that HF currents flow in loops of least inductance.

Dont separate the transformer to far from the caps tough, because long leads may then act as antennas for the current spikes inherent from caps recharging.

 

[dsavitsk]

Quote:

Originally Posted by DaKi][er Ground plane, no reason to not use one

My first try used one. I took it out because I doidn't know what to do with dgnd (though I still don't.) If I can just connect it to the plane then I'll try it again.

Quote:

Originally Posted by DaKi][er also, I wouldn’t have the transformer board mounted either

This seems to be everybody's least favorite part, but I think it has to stay. This is a replacement in a very small space for a guzzler usb dac. There really isn't room to move it anywhere else in the case. So, I agree that it is not ideal to have it on board, but I think there aren't any other options. Also, since this is the expresPCB deal, the board size is constant. Removing it doesn't gain me anything in that regard.

Quote:

Originally Posted by DaKi][er After that it looks ok, I’m no fan of non-os or resistor i/v in either sound quality and in theory behind it but lots of people do like it

My experience is that I do, for whatever reason.

Quote:

Originally Posted by halman You have nowhere near enough clearance between the fuse and the chassies mounting holes.

Hmmm. The case this is going in is wood. Even so, Perhaps if I either shrink or remove the mounting hole. I feel like a fuse is important

and I'm not sure it has anywhere else to go.

Quote:

Originally Posted by halman There are two windings on that toroidal I recomend you dedicate one of them to the Tda1543.

It sounds like a good idea, but I don't know where to put the second set of smoothing caps, etc. Maybe I'll try to work them in, though.

Quote:

Originally Posted by halman And do use a groundplane but remeber that HF currents flow in loops of least inductance.

What are the implicaions of this?

Quote:

Originally Posted by halman Dont separate the transformer to far from the caps tough, because long leads may then act as antennas for the current spikes inherent from caps recharging.

Are they too far as they are?

 

[Clutz]

Quote:

Originally Posted by dsavitsk My first try used one. I took it out because I doidn't know what to do with dgnd (though I still don't.) If I can just connect it to the plane then I'll try it again.

Hello dsavitsk,

Excuse my ignorance - I'm still busy reading AoE, and some online electronics resources. I had a couple of questions and a couple of comments.

First a couple of general questions: What are the overall design constraints? You said that this was supposed to be a replacement for a guzzler DAC, are there any other constraints?

Regarding the ground plane issue, could you make a seperate ground plane for the digital portion, and connect the two together using a ferrite bead a la Monica 2 DAC?

Why not make two seperate power supply units - one for the digital section, and one for the analog section? Is it ecause of space constraints on the board? If so, what about cutting away from a baby-steps like PSU to a simpler pair of PSUs if that would fit...? (I don't mean to presume anything here, I'm just curious and doing my best to learn).

Additionally, I'm curious why you are using the 1543 instead of the 1545?

I'm excited about this project.

Are you planning on posting the schematic?

 

[dsavitsk]

Should the ground plane cover the entire board, or only the DAC circuit portion.

Quote:

Originally Posted by Clutz Could you make a seperate ground plane for the digital portion, and connect the two together using a ferrite beed a la Monica 2 DAC?

I think the only connection to the digital ground is pin 20 of the pcm2707, so it would be a ferite bead between this pin and the rest of the ground plane. The pon still has to conect to the ground directly somewhere though, and I am not sure where that would be that would be any different? There was a pretty heated debate on the usefulness of the ferite bead at diyaudio revently, and my recollection is that the take away point was that people thought it was a bad idea, though I don't think I understood why.

 

[halman]

Only cover the digital/dac portions of the board leave the AC side blank. I would suggest to keep the groundplane solid.

Here are a few references:


http://www.hottconsultants.com/techt...gnd-plane.html

http://www.hottconsultants.com/techtips/tips-slots.html


You may also do a search for guido tents .pdf on supply decoupling.

 

[Clutz]

Quote:

Originally Posted by dsavitsk Should the ground plane cover the entire board, or only the DAC circuit portion. I think the only connection to the digital ground is pin 20 of the pcm2707, so it would be a ferite bead between this pin and the rest of the ground plane. The pon still has to conect to the ground directly somewhere though, and I am not sure where that would be that would be any different? There was a pretty heated debate on the usefulness of the ferite bead at diyaudio revently, and my recollection is that the take away point was that people thought it was a bad idea, though I don't think I understood why.

I hadn't considered just putting a ferrite beed betwen the digital ground and the analogue ground plane. I guess I still have a long way to go, and a lot of things to learn - but that's what keeps it interesting. BTW: I edited my post while you were responding to it - do - so I have many more questions for you if you have the time to answer them.

Thanks,
Clutz

 

[dsavitsk]

Quote:

Originally Posted by Clutz Excuse my ignorance - I'm still busy reading AoE, and some online electronics resources. I had a couple of questions and a couple of comments.

You'll know a lot more than I do at that point. My knowledge comes from reading these boards, data sheets, asking a lot of dumb questions, and soldering stuff together. Sometimes it even works. I also read the NEETS manual which is the best thing I have found for learning.

Quote:

Originally Posted by Clutz First a couple of general questions: What are the overall design constraints? You said that this was supposed to be a replacement for a guzzler DAC, are there any other constraints?

The primary one is size. ExpressPCB has a deal where you can get 3 boards for $60, but they have to be of a specific size. Also, this is intended to fit in a box that is the same size as the RA-1, so things like moving the transformer elsewhere really just aren't an option.

That said, while I am hoping to improve on the quality of Gus's design (which I think is pretty good in its own right), this is not intended to be the end all and be all of quality. It should be pretty good.

Quote:

Originally Posted by Clutz Why not make two seperate power supply units - one for the digital section, and one for the analog section? Is it ecause of space constraints on the board? If so, what about cutting away from a baby-steps like PSU to a simpler pair of PSUs if that would fit...? (I don't mean to presume anything here, I'm just curious and doing my best to learn).

The environment this will be used in is in a neuroscience lab with lots of digital equipment running, so I think the filter on the power line is actually more important than dual supplies. Also, since there are multiple voltage regulators, I am hoping that it'll kind of be like multiple supplies. Anyway, if I remove the filter, I can't call it a baby steps.

Quote:

Originally Posted by Clutz Additionally, I'm curious why you are using the 1543 instead of the 1545?

I have some. I bought a bunch to experiment with running them in parallel, and I have a couple (literally 2) left over.

Quote:

Originally Posted by Clutz I'm excited about this project. Are you planning on posting the schematic?

Yeah, I should. Right now it's in my head. All it is is the 00940 linked schematic with a few decoupling caps + most of the steps but with the lm317 specific stuff and C2 removed + the IV which is a 1K resistor to ground, a cap, then a 47K resistor to ground (+ a 1K bias resistor.) Oh, also the power to the TDA1543 is from peter daniel's verion of the tda1543 dac.

As soon as I am done and things work I'll post the board files too in case anyone wants their own.

 

[halman]

Don´t put a bead in series with the chips DGND, everything should be referenced to the same ground potential, and internal capacitances in the chip will decouple HF currents elsewhere in an uncontrolled (and undesired) fashion.

 

[Clutz]

Quote:

Originally Posted by dsavitsk The environment this will be used in is in a neuroscience lab with lots of digital equipment running, so I think the filter on the power line is actually more important than dual supplies. Also, since there are multiple voltage regulators, I am hoping that it'll kind of be like multiple supplies. Anyway, if I remove the filter, I can't call it a baby steps.

Sounds like we work in similar environments.

Quote:

Yeah, I should. Right now it's in my head. All it is is the 00940 linked schematic with a few decoupling caps + most of the steps but with the lm317 specific stuff and C2 removed + the IV which is a 1K resistor to ground, a cap, then a 47K resistor to ground (+ a 1K bias resistor.) Oh, also the power to the TDA1543 is from peter daniel's verion of the tda1543 dac. As soon as I am done and things work I'll post the board files too in case anyone wants their own.

I think I may give it a try.

 

[dsavitsk]

Quote:

Originally Posted by Clutz Sounds like we work in similar environments.

My wife is the neuroscientist, actually. I'm a J.D. working on a Ph.D. in social science.

 

[dsavitsk]

A few updates. I added a ground plane. The PS ground connects to it at the ground pin of the LM7809. The biggest issue I see is that the power to the TDA1543 cuts a lot of the ground plane in two. I am not sure if this is a problem, and I'm not sure what to do about it.

Otherwise, I just moved a few things around to clean up some stuff. I also didn't deal with the fuse issue yet.

 

[Nisbeth]

Quote:

Originally Posted by dsavitsk The biggest issue I see is that the power to the TDA1543 cuts a lot of the ground plane in two. I am not sure if this is a problem, and I'm not sure what to do about it.

I'd replace the traces with two long jumpers. It may not be neat but it keeps the ground plane intact

/U.