R2R DACs 2017

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
I would say the Aries Cerat Kassandra range is very interesting indeed. The Ref II is about 18K euros and sports probably the best sounding chip ever made, the AD1865NK, but uses 16 of them in a new digital engine. It also has a powerful tube output, I/V stage using custom transformers, and output transformers that result in a line level of 6v or 10v at 18 ohms or 40 ohms. It has a 384 USB input and SPDIF with a super clock. Hugely over specified power supply, it goes on. I am thinking this is my destination DAC, watch this space of my review soon .... I think my 20 year search is going to be over. So many DACs been and gone, mostly NOS and latter were all Audio Note models.

To me, Aries Cerat take a different path to so many others, no cutting corners, no small parts, just design it right, even if it ends up big. It comes in at 60 kilos.

See how big it is, next to Stavros (the designer) with an Audio Note M10 (middle shelf) in the background for scale, which is the size of my previous DAC the Audio Note DAC 5.

View attachment 37959

Well, I can say that I love having it at home...it is last DAC for me (except a higher Aries Cerat perhaps).
 

nonesup

VIP/Donor
Feb 15, 2017
1,044
623
355
Spain
Very interested in this DAC and I would still be more interested, if I had RJ45 input.
 

Legolas

VIP/Donor
Dec 27, 2015
1,042
387
455
France
That is highly suspect/debatable. The TDA1541, PCM63, PCM1704 and UA D20400 chips are all generally held in higher regard...which also explains why they’re much more difficult to source.

Don't agree. The AD1865 NK is also difficult to source I have found (genuine chips). I have modded a few Audio Note DACs and they use this chip as well.

Generally held in higher regard, maybe, maybe not. By who? But I agree, these are the clusters of the best NOS chips IMO. Though know nothing about the UA chip.

One of the aspects to the AD1865 is it has simple logic and is possible to implement and avoid filtering or other digital logic / processes. So you can make it send data as near straight through as possible. I believe the PCM1704 is more difficult to implement in this way.
 

Tam Lin

Well-Known Member
Mar 20, 2011
96
36
923
79
North Texas
The AD1865 NK is also difficult to source I have found (genuine chips).
Maybe you haven't looked in the right places. Rochester Electronics, which specializes in EOL components and acquired all the remaining AD inventory, has thousands in stock.

I believe the PCM1704 is more difficult to implement in this way.
Actually, the PCM1704 is easier to implement, straight through, as you say, than the AD1865.
 

flyer

VIP/Donor
Dec 16, 2012
423
179
1,160
Belgium
Very interested in this DAC and I would still be more interested, if I had RJ45 input.

Aries Cerat allows quite some customizations to its models, implementing BNC or a true balanced output or ... are all possible, some even free of charge. But implementing a RJ45 would this not mean that you expect the music server to be part of (i.e. inside) the DAC?
 

nonesup

VIP/Donor
Feb 15, 2017
1,044
623
355
Spain
I do not pretend that the server is inside the DAC, what I would like is that (in addition to its other USB inputs, BNC ....) could be added an RJ45 rendering input, such as the models of MSB and other manufacturers Lumin , Linn, Brinckman ....
 

Al M.

VIP/Donor
Sep 10, 2013
8,678
4,467
963
Greater Boston
That is highly suspect/debatable. The TDA1541, PCM63, PCM1704 and UA D20400 chips are all generally held in higher regard...which also explains why they’re much more difficult to source.

One chip not difficult to source, but not mentioned, is the AD5791BRUZ. It is current production:

http://www.analog.com/media/en/technical-documentation/data-sheets/AD5791.pdf

and used in the Schiit Yggdrasil DAC.

It is probably hard to say if the AD1865 (from the same company, Analog Devices) or one of the above mentioned chips are better sounding unless a direct comparison is made, with the same type of implementation.

The AD5791 is, however, very difficult to implement because of glitch energy (hence it is not suggested by the manufacturer as audio DAC chip). Mike Moffat (of Theta Digital fame, now co-owner of Schiit) has found a way to address the problem, without sample & hold devices which he says sound bad. What he thinks is a strong point of this expensive precision *) DAC chip is its great linearity (see INL and DNL numbers in datasheet). He says most audio DAC chips feature such bad linearity numbers that they are not even mentioned in their data sheets.

_____________

*) Applications listed by manufacturer:

Medical instrumentation
Test and measurement
Industrial control
High end scientific and aerospace instrumentation
 
Last edited:

joeling

Well-Known Member
Jan 8, 2014
175
51
335
DCS not R to R ?
 

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
Ring Dac, so yes...a special type.

If I remember correctly it is really a 5 bit DAC being run essentially like a sigma/delta DAC. So, it is kind of both and neither.
 

opus112

Well-Known Member
Feb 24, 2016
462
4
148
Zhejiang
A 5 bit DAC run at a high degree of oversampling is very unlikely to be constructed as R2R. I'll be 32 equally weighted resistors for lowest possible glitch energy.
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
One chip not difficult to source, but not mentioned, is the AD5791BRUZ. It is current production:

http://www.analog.com/media/en/technical-documentation/data-sheets/AD5791.pdf

and used in the Schiit Yggdrasil DAC.

It is probably hard to say if the AD1865 (from the same company, Analog Devices) or one of the above mentioned chips are better sounding unless a direct comparison is made, with the same type of implementation.

The AD5791 is, however, very difficult to implement because of glitch energy (hence it is not suggested by the manufacturer as audio DAC chip). Mike Moffat (of Theta Digital fame, now co-owner of Schiit) has found a way to address the problem, without sample & hold devices which he says sound bad. What he thinks is a strong point of this expensive precision *) DAC chip is its great linearity (see INL and DNL numbers in datasheet). He says most audio DAC chips feature such bad linearity numbers that they are not even mentioned in their data sheets.

_____________

*) Applications listed by manufacturer:

Medical instrumentation
Test and measurement
Industrial control
High end scientific and aerospace instrumentation

The specs presented in the datasheet are static - useless for audio purposes. Mike Moffat is reported to use an an Analog Scharc DSP to process the data before the four DAC chips - so we can not consider it as a "pure" R2R implementation. But IMHO what counts is how it sounds!

We can divide the DAC manufacturers in two classes - those who tell us with detail how they exactly implement their DACs and those who let us guess what they do.
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
A 5 bit DAC run at a high degree of oversampling is very unlikely to be constructed as R2R. I'll be 32 equally weighted resistors for lowest possible glitch energy.

Surely. The Vivaldi DAC uses 48 pairs of precision matched equal resistors http://www.monoandstereo.com/2016/01/the-dcs-story-in-depth.html

"The Ring DAC oversamples to about 3MHz, noise-shapes to 5 bits and decodes the binary data to 32 balanced lines. Instead of requiring sixteen high-precision resistors, the Ring DAC drives 32 pairs of standard resistors. In fact early 900 series converters used 32 pairs, from 950 to Debussy used 44 pairs, and now Vivaldi uses 48 pairs.

The key thing is that you turn on different current sources randomly, so any error is not related to any particular signal coming in – you de-correlate the signal from the error,” he explains."
 

opus112

Well-Known Member
Feb 24, 2016
462
4
148
Zhejiang
The specs presented in the datasheet are static - useless for audio purposes.

There's at least one spec buried in there which is dynamic, and doesn't look great for audio on the surface because they've chosen such a low sample rate for it. The SFDR is 100dB typically, but at a 10kHz sample rate and with a 1kHz stimulus tone. The headlined 1uS settling time turns out to be only to 0.02% (equivalent to 12bits accuracy at 1uS). However this is for a large step size (10V), not really representative of an audio application. The small signal settling time is (and this is super interesting) quoted at a different gain and with a different opamp (AD845 for large signal, AD797 for small signal).
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing