Npc Sm5847/sm5865 Dac

[andrei]

I am building a DAC around the latest and greatest NPC chipset - SM5847 digital filter and SM5865 delta-sigma DACs. SM5865's are quite interesting chips - their specs are comparable to if not better than the PCM1792 or AD1955, while the pin-out is much more board-friendly, IMHO.

I thought I'd see what people think before going into production. Attached is the (preliminary) schematics - digital part is straightforward, analog stage is fully differential (THS4131) and is fed from a low-noise voltage reference. The board layout is almost done - fits on 8x10cm board with SMT components mostly.

If people are interested, I can get a few extra boards manufactured when I go into production...

[Õàëÿâùèê]

Hi, andrei

Thanks for schematic.

What is IC6, IC7A/B? It`s shunt regulator or it`s mistake in inscriptions +5/+12?

SM5847 digital filter is really greatest, congratulation!
But about DACs of NCP I do not know. About the previous versions NCP`s balanced DACs there were very many bad responses. It used on CD-players by Pioneer, which marked them as PDM XXXX. Many people changed them on AD and were very pleased. Someone spoke, what even such **** as CS4390, better, than these PDM.


Regards
Halyavshick.

[andrei]

IC6 is a low noise buried zener 5V voltage reference. It's output is low-pass filtered and buffered by IC7, which is a dual low-noise JFET-input single rail opamp. I do indeed intend to bias the opamp (using R14/R16) into sinking the current, making the whole circuit act as a shunt reference. The output voltage should have less than 10uV of noise, if the 12V rail is resonably clean.

Digital supplies for the DAC chips are shunt as well - although since noise there does not matter much, they are simple TL431's.

As far as DAC chips go - SM5865 looks good on paper (THD+Noise -110dB, which is lower than PCM1792 or PCM1738 and about the same as AD1955). That's why I am giving it a try. Of course, NPC makes a whole bunch of cheap consumer DACs as well, which probably are not much different from everybody else's cheap DACs...

[Kovax]

Hi, andrei !

Where can one get that NPC SM5847 digital filter?

Regards, Kovax

[Õàëÿâùèê]

Hi,
andrei
Thank.
It is meaningful and useful to try it in any case, but it is desirable to compare to others DACs IC similar to this.

But, I think, what multibit R-2R will be better. But where to take 24 bit R-2R?

Kovax
Address to support NPC, there you will direct to European distributor TAXAN Ltd
U.K. Bracknell, Berkshire, RG12 1RB
tel. 01344-469830
fax.01344-860981

[andrei]

Oh, I have built a DAC around PCM1730 before (posted here), and lived with a few commercial DACs (including Audio Note), so I have a few data points . But you are right, the whole point of this design is to compare SM5865 against others. The thing I like about SM5865 is the pinout - you can actually do the layout on a single layer, which means good ground plane!

As far as R2R DACs go, for 24-bit the only thing around is PCM1704. I have been thinking about a DAC around it for a while - take a look at the attached schematics (it is a draft, important stuff is not routed yet, but it should give you an idea of what I've been up to)...

To Kovax:

NPC distributors are listed on their website: http://www.npc.co.jp/en/contact/distributors.html
The product list and datasheets are here: http://www.npc.co.jp/en/product/index.html

[andrei]

Oh yes, the schematics...

[Kovax]

Thanks guys for the hints.

Best regards, Kovax

[Õàëÿâùèê]

24 bit at PCM1704 is a beautiful marketing fairy tale.
One mine familiar (he is great professional) measured real linearity (monotony) of 1704-J. As was determined, it is worse than at 20 bit PCM63P/K. And in a sound too there is a distinction to worse. Best R-2R is AD1862, in his opinion.

Thank for schematic. I do not understand, for what some people connect in a parallel R-2R? IMHO, parallel (or differential, that is even worse ) connect it only worsen a sound in a case R-2R. There are no two completely identical R-2R matrixes never.
Subjectively, it results in a stratification of a sound.

Regards.

[andrei]

24-bit DAC just means it can accept 24-bit data, not output it . I am not that surprised at your friend's findings, datasheets pretty much say the same thing:

PCM63P-K -100dB typical THD+N,
PCM1704U-J -96.5dB typical THD+N,
PCM1704U-K -102dB typical THD+N,
AD1862N-J -98dB typical THD+N.

Subjectively, the only DAC with PCM1704 I had (Bel Canto), I did not like at all. But that might not have been the DAC chip; there are plenty of ways to screw up the sound, as I am discovering for myself . Too bad PCM63 and AD1862 are obsoleted by manufacturers...

My reason for differential connection is not to improve linearity, but to allow analog stages with decent common mode rejection. Paralleling DAC chips will improve signal to (uncorrelated) noise ratio, but the gains are small for the amount of trouble you have to go through.