[ble0t]

Quote:

Originally Posted by ezkcdude What's wrong with having GND go through vias? It's not a signal line. How else would SMT components connect to internal GND planes in multi-layer boards, anyway? It seems to me your design will be a lot simpler if you dedicate the bottom side to a large ground plane.

Take a look at this: http://www.tentlabs.com/Info/Article...decoupling.pdf

It is also noted in various datasheets and has been discussed on this forum that you want to have the groundplane directly underneath the DAC and as full as possible...a search on 'groundplane' and 'DAC' netted some results for me concerning that...

 

[ezkcdude]

You want short signal traces and unbroken ground plane. Make that happen, and everything else will fall in place, right?

 

[Garbz]

That is not a very conventional I/V stage you have there. I am not even sure it will sound as good as a standard opamp I/V converter. Typically most DAC chips require a virtual dead short on the output to dump their current. So in essence by running it through a resistor or other passive I/V stage it is often detrimental to the sound. An opamp configured as an I/V converter presents zero impedance to the input and the design is easy which is why it is often recomended in the datasheets.

One thing is certain you should probably add a 8pin header and then use 4 jumpers to connect the opamp I/V stage. That way in a few months if you get board you can built a transimpedance I/V converter and just remove the jumpers and connect it up via a ribbon cable.

 

[t52]

regarding the groundplane on the opposite layer of the dac chip and running gnd hrough vias:
i think the effect we will have is called "ground bounce", at least somebody gave it that name in an article i read some time ago. it will be caused by the small inductances between chip gnd connections and groundplane... i think i'll go for a solid ground connection first and avoid vias there and add some vias on the signal lines instead... the longer i look at the first layout, the better i like it, the groundplane is very solid on the top layer (imho). the power supply/regulators/capacitor placement can be improved quite a bit but i think the chip placement is pretty good. think i will go back to that design and fix the psu section, add a break in the groundplane next to the dacs to have analog and digital portions (as recommended by garbz) and add an extra power input for the analog dac supplies. there should be no remaining traces going across the digital/analog border anymore (except inside the dacs of course).

regarding i/v stage: garbz you're right, if i already have 4 opamps on there i can just as well use them as direct i/v converters and do the summing then, so the dac sees a virtual gnd on it's current outs...
the changing of different i/v stages should not be a problem, because i'll have it on a different board, the "digital" board has a female 25pin subd connector on it and will carry the 8 current outs of the two dacs and the 4 Vref lines...

thanks for all your input, i appreciate it!

 

[00940]

You could also use your summing opamp to add a second RC filter, to get a sharper cut off.

 

[t52]

Quote:

Originally Posted by 00940 You could also use your summing opamp to add a second RC filter, to get a sharper cut off.

hmm, do you think that's necessary? i was told these chips with internal filters don't need that sharp rolloff characteristics anymore.
anyways, do you have a link to a circuit i can have a look at? i'd like to simulate the filter frequency/phase/transient response before implementation...

 

[Garbz]

I used a 2nd order filter in my design. Though I think I stuffed it up because I was missing components and the second curve kicks in at about 100khz. Better late then never I suppose.

Most datasheets I have seen even with very advanced digital filters they still show the "recomended" output stage as multiple opamps normally configured so that the first is an I/V stage with a 1st order lowpass, and the second is setup as a multiple feedback filter which supplies a further 2nd order lowpass response.

I am not sure what this will have to do with the sound quality, but in another thread recently I learnt it has more to do with not blowing up your preamp. As mine still works and is on all day I can only assume it was sufficient. It is probably wise at the least to put a LP filter in the I/V and then a second RC network after it.

 

[mzperx]

Hi All,

Do anybody have 4-6pcs SC947-02 S/PDIF transformer to sale in europe?
Or could you propose sale point except of ScientficConversion from US?

thanks
Zoltán

 

[tubekeeper]

Anybody can send me the DAC 5r5schematic and the pcb layout file to the below e-mail sotam@jetfly.myserver.org

I can't down;oad from the site.

 

[smoothlondoner]

could i construct a separate board for upsampling to intergrate with  this dac?