Non OS TDA1543 PCB Layout

[Gcollier]

Hi all,

I've attached some pics of my PCB layout for a planned TDA1543 DAC. I am using a CS8414 reciever following the standard non-os trend. The schematics are basically the same as Fedde's NONOZ III design: jwg.student.utwente.nl/fedde/nonoz3.html

The plan is to use the bottom of the board as the power ground and the top as the digital ground. You will see in the picture that the power components are not isolated from the digital ground in the top layer. This won't be a problem as I am not using plated through holes since I am etching the boards myself. the two ground planes will connect at the power supply connection.

My plan is to mate this with a PCM2906 or 2706 USB DAC, and an OPA627/Buf634 preamp that I have already completed.

Any comments or suggestions are very welcome, as this is only my second digital design.

G.

[hagtech]

You may want to leave yourself a possible connection between grounds right at the DAC, instead of at power supplies. Separate grounds can be a very dangerous thing, having exactly the opposite effect intended.

Look at the return current paths for every signal. My concern here are the digital lines between receiver and converter. Just where does the return path go? You have these very fast transient edges going direct from one chip to the other, then a return current going way out to the edge of the board and back. This is a discontinuity and forms a large inductive loop and crosstalk. There are also connections within the converter chip itself (it helps to have a complete schematic of the innards of each chip) passing transient signals from one ground system to another. These transients have no local return path, hence inducing glitches right where you thought you got rid of them.

IMHO, the best place to connect different ground & supply systems is right where the two meet, under the converter chip.

jh

[jean-paul]

Some suggestions: Implement the option for an input transformer. Edit the layout so that film caps can be used at the output. Break the loop in the groundplane above the input caps. Add some extra SMD caps right at the supply pins of CS8414. Maybe you can add the sin x/x rolloff filter ??

Personally I would not design the board for SMD resistors only as normal metal film ones sound better and availability is no problem.

Looks good !

[Gcollier]

Jean-Paul/Hagtech

Wow, thanks so much for the suggestions. This is the kind of help I was looking for. I'm quite the noob when it comes to digital design...I'm sure it shows.

Hagtech, as for the connection between the grounds, shoul this be at the supply from the DAC (this is already pretty much the central ground point, or should it be at the output? I will make an attempt to reduce the size of the groundplane to shorten the return paths as much as possible.

Jean-Paul, could you be a bit more specific as to the "loop in the groundplane above the input caps". I'm looking but I don't see it??? I do se that I broke the power ground at the SPDIF input, I will fix this later tonight.

What are your thoughts on star grounding. Would I really be better off with just a single ground rather than making any attempt to keep them seperate?

Thanks

G.

[hagtech]

I haven't looked at the layout ... But in general, if you choose to use two ground sections, tie them together at the DAC. Not at output connectors or power supply connectors.

Start grounding is great, if done right. Not as simple as it sounds. ALL signals and ALL supplies must use the same star. Otherwise benefits are lost. Makes things more difficult.

jh

[tvr2500m]

Where's a good source/s for the TDA1543 and 1541 chips?

- SJ

[Guido Tent]

Originally posted by hagtech

Look at the return current paths for every signal.
jh

Yes, yes, yes ! Think in current loops guys, Kirchoff must have an internet connection somewhere, so watch out .......

http://www.tentlabs.com/Info/Article...decoupling.pdf

best

[tubee]

May a good thing to ask here:

My CDP has a groundplane (Philips CD660 and others), and the circuit uses several conecting points on that groundplane to connect parts to gnd, these points are spreaded on several places of whole groundplane. My idea is to isolate those connections, and choose ONE star-point (near dac) and connect all the ground conections from there with massive 1.5 gauge wire.
Wil this improve grounding layout?

Thanks!

[Guido Tent]

Originally posted by tubee
May a good thing to ask here:

My CDP has a groundplane (Philips CD660 and others), and the circuit uses several conecting points on that groundplane to connect parts to gnd, these points are spreaded on several places of whole groundplane. My idea is to isolate those connections, and choose ONE star-point (near dac) and connect all the ground conections from there with massive 1.5 gauge wire.
Wil this improve grounding layout?

Thanks!

Hi

No it won't, someone suggested to look at the current loops.....

Attaching them to several places to the plane isn't a problem as long as no big (RF) currents flow through the plane. To avoid that, several suggestions are given in my article.

best

[Gcollier]

Alright, I decided to try and star ground this thing. I also opted to use through hole resistors rather than SMD. Not because of the quality point, but because it makes it easier to couple traces between top and bottom.

You will see that the central ground is at the DAC. The routing gets a bit messy, so if you can see a way to shorten some of the traces let me know. I tied AGND and DGND of the CS8418 together beneath the chip...will this present a problem or do I need to keep them seperate until they meet at the central ground point?

The dimensions of the board are 2.5" x 2.25"

Anxiously awaiting your suggestions

G.