Hi,
I got 4 PCM1704 ridiculously cheap and put together the above components just following data sheets and evaluation board descriptions.
I use Nelson Pass' D1 output stage for the I/U-conversion.
The thing sounds incredible as it is, but as I just built it on experimenting board (yes I found one for the DF too!) and connected everything with wires I think there is much room for optimization. So my questions are:
1. What do I win by etching a PCB?
2. Does a quartz oscillator help and if so, which frequency at which part?
(I don't have one yet, the receiver makes the master clock by itself.)
The eval. board describes a 24.576MHz osc. for the DF1704, but if I understood it right, the DF has an oscillator built-in, so that wouldn't help much?
The elektor DAC 2000 (which uses same parts) has a 6.xxx MHz osc. at the CS8414.
Which is the better way?
(I want to play CD and DVD-Audio, so it should be compatible to all sampling frequencies.)
3. Maybe somebody experimented with the slow vs. fast rolloff-filter settings of the DF1704?
4. Power supply. I use a 2x15V transformer, one 7815 / 7915 pair, one
7805/7905 air for the DACs and one 7805 each for filter and receiver.
(output stage is completely separated)
I guess here is a lot of room for improvement.... Is there a schematic with a high-quality solution somewhere?
5. That's all I can think of for now, further suggestions are very welcome
thanks,
martin
I got 4 PCM1704 ridiculously cheap and put together the above components just following data sheets and evaluation board descriptions.
I use Nelson Pass' D1 output stage for the I/U-conversion.
The thing sounds incredible as it is, but as I just built it on experimenting board (yes I found one for the DF too!) and connected everything with wires I think there is much room for optimization. So my questions are:
1. What do I win by etching a PCB?
2. Does a quartz oscillator help and if so, which frequency at which part?
(I don't have one yet, the receiver makes the master clock by itself.)
The eval. board describes a 24.576MHz osc. for the DF1704, but if I understood it right, the DF has an oscillator built-in, so that wouldn't help much?
The elektor DAC 2000 (which uses same parts) has a 6.xxx MHz osc. at the CS8414.
Which is the better way?
(I want to play CD and DVD-Audio, so it should be compatible to all sampling frequencies.)
3. Maybe somebody experimented with the slow vs. fast rolloff-filter settings of the DF1704?
4. Power supply. I use a 2x15V transformer, one 7815 / 7915 pair, one
7805/7905 air for the DACs and one 7805 each for filter and receiver.
(output stage is completely separated)
I guess here is a lot of room for improvement.... Is there a schematic with a high-quality solution somewhere?
5. That's all I can think of for now, further suggestions are very welcome
thanks,
martin
I would really encourage you to leave out the input receiver together with the digital filter i.e. to go the non-oversampling filterless route. Try to connect these PCM1704's to the EIAJ or I2S output of your player (decoder) directly. At least do yourself a favour and try it once before you start fiddling around with clocks, filter slopes, power supplies and complicated PCB designs. Please!
yes I would probably like to try this although what I have now sounds better than anything I heard before.... 
I would have to find the I²S connections in my cd/dvd-audio player and convert them to a format suitable for the PCM1704. I think these are 4 lines? Lots of work...
Is there a finished concept about this? How do I get left and right SDATA signals without the DF?
I would have to find the I²S connections in my cd/dvd-audio player and convert them to a format suitable for the PCM1704. I think these are 4 lines? Lots of work...
Is there a finished concept about this? How do I get left and right SDATA signals without the DF?
Martin, you win a lot by puting it on a PCB. Most importantly, you put yourself in the driver seat when it comes to board partitioning, location of return currents (important if you're wireing together demo boards), grounding, and decoupling. Search for "Supply decoupling and layout of circuits with digital ICs" by Guido Tent. I got it from this forum.
I've used Eagle a lot with these chips. Beware that the footprint of the PCM1704 isn't straight out of the box. If you go for a board, remember to print it out and try to fit the chip! Are you using Eagle? If so I can mail you some of my stuff including the PCM1704.
What kind of analog filter are you using? I was never one of these non-OS guys, rather I go for custom analog filters and digital oversamplers born in Matlab and implemented in Xilinx.
I clock the DACs from the crystal oscillator in the CD drive itself. It is this clock (multiplied or divided) that is multiplexed into the SPDIF stream and extracted by the CS8414. On my current boards I can select between CD drvie clock and CS8414 derived clock. I don't let an oversampler handle clocking because the 384x or 256x clocks I get from CD drive or CS8414, respectively, make nice frequencies for a bitclock with 24bits at 8x oversampling. I actually use either one directly to the bitclk on the DAC chips and make the FPGA run asyncronously by polling it.
Greetings,
Børge
I've used Eagle a lot with these chips. Beware that the footprint of the PCM1704 isn't straight out of the box. If you go for a board, remember to print it out and try to fit the chip! Are you using Eagle? If so I can mail you some of my stuff including the PCM1704.
What kind of analog filter are you using? I was never one of these non-OS guys, rather I go for custom analog filters and digital oversamplers born in Matlab and implemented in Xilinx.
I clock the DACs from the crystal oscillator in the CD drive itself. It is this clock (multiplied or divided) that is multiplexed into the SPDIF stream and extracted by the CS8414. On my current boards I can select between CD drvie clock and CS8414 derived clock. I don't let an oversampler handle clocking because the 384x or 256x clocks I get from CD drive or CS8414, respectively, make nice frequencies for a bitclock with 24bits at 8x oversampling. I actually use either one directly to the bitclk on the DAC chips and make the FPGA run asyncronously by polling it.
Greetings,
Børge
Sorry, I wasn't aware that the PCM1704 is not stereo and you will definitely need some glue logic. It also depends on the format that is available in your player. Most probably there's no finished and proofed concept but there are similar examples. One is here: http://www.diyaudio.com/forums/showthread.php?s=&threadid=39993
Still, try to get around the input receiver and the DF once, maybe with another DAC and/or player...
Borge, thanks for the paper hint, I will need this if I ever make a PCB...
I think I would use Eagle, it would be my first PCB, I did everything on experimenting board until now... So your Eagle stuff would probably help me.
I don't use any analog filter that I am aware of... the D1 output stage has 2 "1st order low pass filter" capacitors on in- and output and I even took them out because I felt they took away too much.
I think I can hear some aliasing but of course I don't know where it comes from. This is why I am thinking of optimizations.
You have a separate line for the clock from your CD drive?
I guess this is only possible with drive and DAC in the same case?
I would like to have the DAC independent of the drive/cd player.
Or would it be possible to use a computer DVD drive as DVD Audio player? I saw some people using computer CD drives but no DVD yet - ?
m.
I think I would use Eagle, it would be my first PCB, I did everything on experimenting board until now... So your Eagle stuff would probably help me.
I don't use any analog filter that I am aware of... the D1 output stage has 2 "1st order low pass filter" capacitors on in- and output and I even took them out because I felt they took away too much.
I think I can hear some aliasing but of course I don't know where it comes from. This is why I am thinking of optimizations.
You have a separate line for the clock from your CD drive?
I guess this is only possible with drive and DAC in the same case?
I would like to have the DAC independent of the drive/cd player.
Or would it be possible to use a computer DVD drive as DVD Audio player? I saw some people using computer CD drives but no DVD yet - ?
m.
Elso Kwak said:
Connecting 24 bit PCM1704 DACs to a 16 bit format will result in a pretty low volume sound as the most significant bits are not used. You need some kind of shift register to get this right but why using a 24 bit DAC on a 16 bit format, NON-OS?
It will have exactly the same volume as when it is fed with 24bit data. You have to align the MSB of the 16bit data stream with the MSB of the DAC input. A working basic circuit can be found by clicking the link provided by Hattori Hanzo. Modestly said, I think the PCM1704 is a good alternative to TDA 16bit DACs.
dieringe said:
Yes I've compared slow vs. fast roll of and I prefered slow filter. Switching the filter response is quite easy to implement. Just compare and see what's more apealing to your ears.
As you've already removed the filter from your D1 outputstage, I guess you will like the soft roll of setting. Removing the whole digital filter is like going one step further.
Re: Re: Optimizing a CS8414 + DF1704 + 4x PCM1704 DAC?
Do you really advise to remove analog and digital filtering? At least one has to be careful about that, as it throws lots of garbage above 20kHz to the amp+speaker. The output of a D/A converter is stepped (settling time of 1704: 200ns), with resulting high frequency content. A 200ns step gives me 2.5MHz......
Ciao, Tino
Fabian said:
Do you really advise to remove analog and digital filtering? At least one has to be careful about that, as it throws lots of garbage above 20kHz to the amp+speaker. The output of a D/A converter is stepped (settling time of 1704: 200ns), with resulting high frequency content. A 200ns step gives me 2.5MHz......
Ciao, Tino
Yes, as a proposal to answer some (all?) of dieringe's questions. Compared to his intention (receiver, DF, clock, PCB etc.) this would be an easy and worthwhile experiment. I cannot really comment on the DF1704 though.
Maybe because it might sound good?
Yes, and also about the various debates regarding the filterless versus filtered approach. I didn't say the filterless way sounds best under all circumstances but I would at least give it a try. And it doesn't hurt...
Re: Re: Optimizing a CS8414 + DF1704 + 4x PCM1704 DAC?
well my perception is that the fast roll-off gives a little better resolution, the slow roll-off gives a more analog-like sound. can't really decide which I like better.
I have another question:
6. there is a bunch of capacitors at each PCM1704. As this is HF stuff I guess tantalum electrolytics are the preferred type?
m.
Fabian said:
well my perception is that the fast roll-off gives a little better resolution, the slow roll-off gives a more analog-like sound. can't really decide which I like better.
I have another question:
6. there is a bunch of capacitors at each PCM1704. As this is HF stuff I guess tantalum electrolytics are the preferred type?
m.
Re: Re: Re: Optimizing a CS8414 + DF1704 + 4x PCM1704 DAC?
Here's what Madrigal's using in their $17K DAC . The chips under nude Vishays are PCM1704 http://www.marklevinson.com/image_library/30_6DAC_lo.jpg
dieringe said:
Here's what Madrigal's using in their $17K DAC . The chips under nude Vishays are PCM1704 http://www.marklevinson.com/image_library/30_6DAC_lo.jpg
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.