Hi,
I'm actually trying to design a DAC based on TI component, and specialy pcm1730 DAC.
Why this pcm1730 ? because I could get it as sample by TI, in replacement of my previous idea (the pcm1704).
I would like to use several pcm1730 in parallel to reduce the S/N ratio, but I've a problem to make an asymetric output (based on AOP - no tube for the moment), due to the fact that the pcm1730 is using a common VREF for output (calles VCOM2), and also that I'm a newbie in the domain.
Could you give me some advise or idea? Did someone of you realized already such output stage ? 😎
Thanks
Xavier.
I'm actually trying to design a DAC based on TI component, and specialy pcm1730 DAC.
Why this pcm1730 ? because I could get it as sample by TI, in replacement of my previous idea (the pcm1704).
I would like to use several pcm1730 in parallel to reduce the S/N ratio, but I've a problem to make an asymetric output (based on AOP - no tube for the moment), due to the fact that the pcm1730 is using a common VREF for output (calles VCOM2), and also that I'm a newbie in the domain.
Could you give me some advise or idea? Did someone of you realized already such output stage ? 😎
Thanks
Xavier.
I think PCM1730´s S/N ratio is low enough, I would not think that you´ll hear the difference if the analog conditioning isn´t the worst after the DAC...
PCM1738 offer´s you the possibility of mono mode where you´ll get about 3db better results, but it needs a µC..)
PCM1738 offer´s you the possibility of mono mode where you´ll get about 3db better results, but it needs a µC..)
Current output DAC's parallell easily. All you have to do is verify that you set up the references correctly (of connected together, reduce the value of the sense resistor by n, if not connected, keep normal. Then of course you have to set up any op-amps in the output filter/gain stage for correct gain.
Petter
Petter
Hi Petter,
Thanks for your reply, but I don't understand your suggestion.
As output of the pcm1730, you have R+,R-, VCOM2, L+ and L-
If I take 2 pcm1730, I'll have 2 different VCOM2, and therefore, I'll have separeted I/V stage fo each R and L output.
Or should I forget the VCOM reference voltage ?
Do you have any picture to illustrate your idea ?
Thanks.
Xavier.
Thanks for your reply, but I don't understand your suggestion.
As output of the pcm1730, you have R+,R-, VCOM2, L+ and L-
If I take 2 pcm1730, I'll have 2 different VCOM2, and therefore, I'll have separeted I/V stage fo each R and L output.
Or should I forget the VCOM reference voltage ?
Do you have any picture to illustrate your idea ?
Thanks.
Xavier.
VCOM2 is just a reference to enable relatively easy setting of the resulting output voltage to zero by setting up a reference at the voltage point at which the current output resides. This sets up a common mode voltage which is nulled by the output stage.
What I propose you do:
1. Have Iref resistor and short the IREF's between the two devices.
2. Short R1+ to R2+, R1- to R2- etc.
3. Use any one of the VCOM2's
4. Set up correct gain of output stage (probably half of what you have now given that you have twice the output current.
Petter
What I propose you do:
1. Have Iref resistor and short the IREF's between the two devices.
2. Short R1+ to R2+, R1- to R2- etc.
3. Use any one of the VCOM2's
4. Set up correct gain of output stage (probably half of what you have now given that you have twice the output current.
Petter
What exacly is the benafit to this?
What should the theoretical benafit actuall be?
I know arcam parallels four wolfsons in thier FMJ player but why?
If this results in much better sound quality then Ill alter my design to include another two chips, for Balanced operation with two chips per channel. Im going for the best sound quality really with this so if it really is a good thing to do then I will do it. Im usin the PCM1794 BTW. Adding another two chips to the PCB wont be too hard as I can just copy and paste most of it in.
Im thinking of redesigning the whole thing anyway using another program.
Cheers for any help Matt.
What should the theoretical benafit actuall be?
I know arcam parallels four wolfsons in thier FMJ player but why?
If this results in much better sound quality then Ill alter my design to include another two chips, for Balanced operation with two chips per channel. Im going for the best sound quality really with this so if it really is a good thing to do then I will do it. Im usin the PCM1794 BTW. Adding another two chips to the PCB wont be too hard as I can just copy and paste most of it in.
Im thinking of redesigning the whole thing anyway using another program.
Cheers for any help Matt.
5th, is there a source in Europe already for the PCM1794? TI lists no stock, but they give Digikey as a source.
Paralleling outputs the way Petter described might improve S/N a bit and THD even less so.
The classic reason to parallel is that you want to cancel distortion by using both channels on one single stereo chip (because presumably they have better tracking if they are on the same chip) in mono mode , which involves feeding the left signal to chip 1, making sure the the right channel of this chip sees an inverted signal. The right signal then has to go to chip 2, again with the signal inverted for the right channel.
If the chip doesn't have a provision for mono mode built in, this involves rather complicated data separation logic.
Paralleling outputs the way Petter described might improve S/N a bit and THD even less so.
The classic reason to parallel is that you want to cancel distortion by using both channels on one single stereo chip (because presumably they have better tracking if they are on the same chip) in mono mode , which involves feeding the left signal to chip 1, making sure the the right channel of this chip sees an inverted signal. The right signal then has to go to chip 2, again with the signal inverted for the right channel.
If the chip doesn't have a provision for mono mode built in, this involves rather complicated data separation logic.
Ok well im going to use the Mono mode of the PCM1794 anyway so I guess thats not important. The distortion of the chips looks to be very low as it is so that seems even less important again. Well that makes everything easier to make then with a lower parts count.
Cheers Matt.
Cheers Matt.
Hello Xav,
Motivated , It's not a DAC like that which you seek
http://www.anidian.com/audio/pdf/dac_block.pdf
@+
Seb
😉
Motivated , It's not a DAC like that which you seek
http://www.anidian.com/audio/pdf/dac_block.pdf
@+
Seb
😉
Petter > Thanks for the explanation. It's more clear for me now.
Unfortunetly, there is no mono mode on the pcm1730 (only on pcm1738 that require a micro processor, too complicated for me.....)
As I got the DAC as sample, I'll try (in stereo mode) and see. My idea is to use up to 4 DAC in parallel, as a lot of persons done already with tda154x DAC. 🙄
Cheers.
Xavier.
Unfortunetly, there is no mono mode on the pcm1730 (only on pcm1738 that require a micro processor, too complicated for me.....)
As I got the DAC as sample, I'll try (in stereo mode) and see. My idea is to use up to 4 DAC in parallel, as a lot of persons done already with tda154x DAC. 🙄
Cheers.
Xavier.
Why parallell: Less noise etc. There is an evaluation board for PCM1704 which uses several DAC chips in parallel (but it is a bit costly). Look there for inspiration.
re 1730: Why don't you go to 1794 immediately instead of messing around with the 1730????? I am pretty sure the 1794 would be way better than the 1730 stuff you are trying to set up.
Re mono mode: 1794 in mono mode is ever so slightly better than two 1794 in stereo in parallell. The simplest way to go is a bunch of 1794 in parallel, or one in mono mode which is essentially two in parallell + some added benefit of internal cancellations.
Petter
re 1730: Why don't you go to 1794 immediately instead of messing around with the 1730????? I am pretty sure the 1794 would be way better than the 1730 stuff you are trying to set up.
Re mono mode: 1794 in mono mode is ever so slightly better than two 1794 in stereo in parallell. The simplest way to go is a bunch of 1794 in parallel, or one in mono mode which is essentially two in parallell + some added benefit of internal cancellations.
Petter
For sure the pcm1794 looks better on the paper, but it's not the same price as the pcm1730.
By TI, the price of the pcm1794 is nearly the same as a pcm1704U (~13 USD)
In comparaison the pcm1730 cost 4.8 USD...
Therefore, with a pcm1730, I'll not reach a high quality but I should reach a very good ratio quality/price
Perhaps I'll use better component in a 2nd release of my DAC, after I got some experience with the first one....😎
By TI, the price of the pcm1794 is nearly the same as a pcm1704U (~13 USD)
In comparaison the pcm1730 cost 4.8 USD...
Therefore, with a pcm1730, I'll not reach a high quality but I should reach a very good ratio quality/price
Perhaps I'll use better component in a 2nd release of my DAC, after I got some experience with the first one....😎
Cost of $11 vs. $4 should not be an issue, regarding the price of decent caps, resistors and op amps.
The question is more how to source the 1794.
I doubt very much that further reduction of noise on devices as good as the 1704 or 1794 will provide any benefit. If at all, getting rid of distortion, especially in the 5 - 20 kHz might have an audible benefit. This will necessarily involve inverting otherwise identical (same chip or at least same wafer) devices....
The question is more how to source the 1794.
I doubt very much that further reduction of noise on devices as good as the 1704 or 1794 will provide any benefit. If at all, getting rid of distortion, especially in the 5 - 20 kHz might have an audible benefit. This will necessarily involve inverting otherwise identical (same chip or at least same wafer) devices....
Hmmmm improve distortion in the 5-20khz range. The distortion of the PCM1794 is already really impressive anyway as low as most pre and power amps that ive seen so I doubt that lowering it any more would really make a difference as its in the noise floor as it is.
- Status
- Not open for further replies.