How does ring dac work ?

Bernhard · 16th January 2006, 02:12 PM

phase_accurate · 16th January 2006, 02:34 PM

As far as I know it is an arrangement of several similar DACs that are each enabled (or selected if you prefer that term) on a per-sample basis. The choice is made by a random generator in order to randomise linearity errors and the like.

Regards

Charles

rfbrw · 16th January 2006, 02:36 PM

Read the patent.

davidsrsb · 16th January 2006, 03:13 PM

A very simple but ingenious solution.

To make a 4 bit chord dac you need from 0 to 15 identicle current sources. They actually have more than 15 and if the required number is n and different combination of n is chosen every time. This has the effect of averaging the errors out.

Bernhard · 16th January 2006, 05:45 PM

Quote:

Originally posted by davidsrsb
A very simple but ingenious solution.

To make a 4 bit chord dac you need from 0 to 15 identicle current sources. They actually have more than 15 and if the required number is n and different combination of n is chosen every time. This has the effect of averaging the errors out.

But how many current sources you need for 16 bit DAC ? 2^16 +1 ?

Petter · 17th January 2006, 01:04 AM

I seem to remember finding a pretty decent description on the Arcam site. As I recall (I really should go check, but I did not find it even after 10 minutes, so here goes from memory - subject to errors).

There were 5 current sources (I think) which would each be switched on and off to generate an average current out of each one.

There was also a randomizing process which added randomness to which samples would go to which current source.

Your question as to how many current sources you need to generate an n-bit signal is easy to answer. It is a function of how many levels you have (in this case two, zero and one) and how fast you are prepared to transition between the two. It is also a function of the frequency you are averaging at since at any one time, a single such switch will be either zero, one or in transition.

So, not knowing what DCS did, I think Arcam used 5 single bit DAC's with randomizing to minimize effects of mismatch.

Petter

davidsrsb · 17th January 2006, 01:21 AM

The DCs RingDAC is a compromise between bitstream type and classic multibit. It is still a sigma-delta but has a few bits of "real" D/A.

Building a 16 bit ringdac would require over 64k current sources and some incredible control logic.
It also needs oversampling to push what is basically dither outside of the audio band.

Petter · 17th January 2006, 01:32 AM

My previous posting may well be correct as far as Arcam is concerned and their implementation of DCS Ring DAC, but I went the extra mile and checked out the DCS site and they claimed 5 bits at 3MHz. No mention was made of the modulation method which I assume to be PWM with discrete steps. One way to get to 5 bits is to set up 32 currents to switch on and off.

DSD is 1 bit at 2.8MHz. If you had access to the data stream and fed it into your pre-amp directly, you would have the sound since you would hear the average as your system would not respond to the 2.8MHz square wave with odd transitions directly.

CD is 16 bit at 44.1KHz which is typically upsampled to 20 or 24 bits at 8 times that frequency at playback to try to go beyond the original data.

I have some WMA material which is 7.1 channels at 24/192 lossless

which will be interesting to hear once I have a 7.1 setup (which may well be never

as I am contemplating 4.0 instead)

There are different advantages/disadvantages with each approach and a quick search on the web will reveal some of these tradeoffs.

Petter

16th January 2006, 02:12 PM	#1
Bernhard diyAudio Member Join Date: Apr 2002 Location: Munich	How does ring dac work ?

17th January 2006, 01:32 AM	#8
Petter diyAudio Member Join Date: Jan 2001 Location: Scandinavia	So I had to go and look some more ... My previous posting may well be correct as far as Arcam is concerned and their implementation of DCS Ring DAC, but I went the extra mile and checked out the DCS site and they claimed 5 bits at 3MHz. No mention was made of the modulation method which I assume to be PWM with discrete steps. One way to get to 5 bits is to set up 32 currents to switch on and off. DSD is 1 bit at 2.8MHz. If you had access to the data stream and fed it into your pre-amp directly, you would have the sound since you would hear the average as your system would not respond to the 2.8MHz square wave with odd transitions directly. CD is 16 bit at 44.1KHz which is typically upsampled to 20 or 24 bits at 8 times that frequency at playback to try to go beyond the original data. I have some WMA material which is 7.1 channels at 24/192 lossless which will be interesting to hear once I have a 7.1 setup (which may well be never as I am contemplating 4.0 instead) There are different advantages/disadvantages with each approach and a quick search on the web will reveal some of these tradeoffs. Petter

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
DIY faraday ring.	speekergeek	Multi-Way	35	21st December 2003 05:07 PM
Ring any bells?	Lisandro_P	Everything Else	0	24th February 2003 03:41 AM
about the shorted ring (flux stabilizer ring)	haggy	Multi-Way	3	4th January 2002 02:04 AM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

16th January 2006, 02:34 PM	#2
phase_accurate diyAudio Member Join Date: May 2002 Location: Switzerland	As far as I know it is an arrangement of several similar DACs that are each enabled (or selected if you prefer that term) on a per-sample basis. The choice is made by a random generator in order to randomise linearity errors and the like. Regards Charles

16th January 2006, 02:36 PM	#3
rfbrw diyAudio Member Join Date: Oct 2001 Location: .	Read the patent.

16th January 2006, 03:13 PM	#4
davidsrsb diyAudio Member Join Date: Dec 2005 Location: Kuala Lumpur	A very simple but ingenious solution. To make a 4 bit chord dac you need from 0 to 15 identicle current sources. They actually have more than 15 and if the required number is n and different combination of n is chosen every time. This has the effect of averaging the errors out.

17th January 2006, 01:04 AM	#6
Petter diyAudio Member Join Date: Jan 2001 Location: Scandinavia	I seem to remember finding a pretty decent description on the Arcam site. As I recall (I really should go check, but I did not find it even after 10 minutes, so here goes from memory - subject to errors). There were 5 current sources (I think) which would each be switched on and off to generate an average current out of each one. There was also a randomizing process which added randomness to which samples would go to which current source. Your question as to how many current sources you need to generate an n-bit signal is easy to answer. It is a function of how many levels you have (in this case two, zero and one) and how fast you are prepared to transition between the two. It is also a function of the frequency you are averaging at since at any one time, a single such switch will be either zero, one or in transition. So, not knowing what DCS did, I think Arcam used 5 single bit DAC's with randomizing to minimize effects of mismatch. Petter

17th January 2006, 01:21 AM	#7
davidsrsb diyAudio Member Join Date: Dec 2005 Location: Kuala Lumpur	The DCs RingDAC is a compromise between bitstream type and classic multibit. It is still a sigma-delta but has a few bits of "real" D/A. Building a 16 bit ringdac would require over 64k current sources and some incredible control logic. It also needs oversampling to push what is basically dither outside of the audio band.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)