DAC designs

SimonC · Post by **SimonC** » Thu Mar 13, 2008 12:58 pm

One 1543 DAC that looks to be well though out and developed is the DDDAC 1543 Mk2

http://www.dddac.de/

I've not heard one of these, but Doede is a diyaudio regular and has put a huge amount of time and effort into the Mk2, plus he explains the design in a way simple enough that even I can understand.

1541's have a large following, but are difficult and very expensive to get hold of now. Having heard Ian's Cambridge DAC at my place I can now tell why they generate so much fuss.

Maybe we should be casting our nets further afield and look into some of the new chips on the market?

Simon C

Post by **Nick** » Thu Mar 13, 2008 1:16 pm

The DAC in HiFi world used 16 1543s. I suspect its very similar to the DDDAC

pre65 · Post by **pre65** » Thu Mar 13, 2008 1:40 pm

Hi-having a bit of a play while on lunch break

with the CD,internal DAC vs Shek DAC.

To be honest the internal DAC seems to have the edge on first listening !

The volume is a "gnats" more and the sound is slightly "fuller" with a slight increase in bass.

My Cd player is a Cambridge Audio D500 SE so its hardly "state of the art",must dig out my old Yamaha CDX810 tonight and repeat the experiment.

What interests me now is how much benefit is there in a "better" transport,perhaps a dedicated one.I have been looking for a NEC multispin 6X (CDR602) standalone CD player with RCA and digital output but they seem thin on the ground except in USA,and then the freight charge kills the deal.

ed · Post by ed » Thu Mar 13, 2008 2:22 pm

not sure this is entirely relevant but the subject was flogged good and hard around 2000-2001 by tnt.

For those that havn't seen it, the tnt write up of the diy dac by Giorgio Pozzoli is a brilliantly well written treatise(imo), just google for convertus.

I bought a dac-ah in 2006 in preference to a beresford purely on the tnt review as I knew nothing about the wee timourous beasties.....I still know next to nothing about the comparitive merits but I am still more than happy with mine.

I think I was swayed by a comment that said something like......you can't buy the parts for less than the price of the off-the-shelf unit.

Ed

Tony Moore · Post by **Tony Moore** » Thu Mar 13, 2008 2:30 pm

Hi Philip,

In my opinion the transport does make a huge difference. There seem to be so many factors, some of which can be offset by fancy clocking in the DAC, tent link, etc. Some problems are probably just down to the drive itself, errors (both data and time (jitter) related), noise on the power supply from the sledge motor interfering with the signal, etc, etc.

I have found that even with the same DAC, different transports can give varying results. The best I found so far out of several I have tried was the CDPro2 mech which is not really surprising as it's design/build quality is well known. Even better to use I2S and cut out the s/pdif altogether.

Personally I can't see that cheap computer CDROM drives will be manufactured with the same design criteria (for audio extraction) as something like the CDPro2.

Of course, why not bin the transport altogether and use a SqueezeBox!

Then you're more or less guaranteed that your data is going to be accurate and with reclocking you can go as far as you want to reduce the time related errors. (jitter) It may even be possible to use I2S from the SB3 straight into 1543 or 1541 DACs and so potentially the ultimate architecture, if implemented well.

Cheers,
Tony

Post by **Nick** » Thu Mar 13, 2008 2:48 pm

Yes, some of the threads I have read relating to the fiting of pulse transformers to SB SP/DIFF outputs has made me realise its very easy to see how cables could matter, and why the "its digital, so what could go wrong" argument is flawed on so many levels.

And this ignores any drive induced jitter or data read errors.

On the face of it, using i2s looks like a much more sensible transport as it contains a TTL level bit and word clock.

Tony: I was looking more at the problem of the bit order from the SB, the circuit you linked to looks (from others comments) to have some propigation delay issues. But I can't see why a less elegant solution involving a bunch of parallel out and parallel loading shift registers couldn't allow any bit order needed?

Tony Moore · Post by **Tony Moore** » Thu Mar 13, 2008 3:31 pm

Hi Nick,

Yes, that cct is a compromise in some ways, to be simpler. I think John Swenson originally did his own interfacing using a CPLD. I haven't any experience of the coding or use of CPLDs so as soon as I saw that I was put off a bit. Then along came this simpler suggestion which John put together for RioTubes on the SD forum.

My own attempt used a series of shift registers to try to offset the words but I gave up on that after it got way too complicated and it just made horrid noises through the DAC!

It may have been simply too much cross-talk and noise in my cct since I built it on vero. Again, a decent pcb and SMD chips could make it all work ok.

There certainly seems to be a market for it. I can't really see why SD/Logitech haven't embraced external DACs more fully with the new SB Receiver and offered a selectable I2S format output and decent sized pcb pads on which to pick it up from. Maybe they think it would impact on their Transporter sales?

Cheers,
Tony

Post by **Nick** » Thu Mar 13, 2008 3:45 pm

I had a look about the SB site, and it looks as if the source for the firmware is closed, at least I couldn't find it.

And of course a entire new set of problems/features may emerge when the new duet starts to ship.

Tony Moore · Post by **Tony Moore** » Thu Mar 13, 2008 5:03 pm

Hi Nick,

Yes, no-one has managed to hack the SB3 to my knowledge. Also the I2S stream comes from the CPLD chip I think. So it's probably hard coded in the logic.

The new Duet's "SqueezeBox Receiver" seems to be very similar to the SB3 but without the display. There must be enhancements in the firmware to support the remote-only setup but apart from that I don't think there are other differences, EXCEPT that it uses a different DAC chip. I haven't looked at whether the new DAC chip uses the same I2S as the 1541/1543. The DAC in the SB3 could be set to use either format, SD chose to use the easiest to implement I guess. By looking at the new DAC and it's connections it should be possible to see if this one is also set to the same format, I'd suspect it is though.

I asked these questions on the SD BB but got no responses. In fact I've never got a response from there...perhaps they all think I'm mad?

Cheers,
Tony

Post by **Nick** » Thu Mar 13, 2008 5:31 pm

By looking at the new DAC and it's connections it should be possible to see if this one is also set to the same format, I'd suspect it is though.

Unless the setting are controlled from the process instead of being hard wired, then its not obvious

Greg · Post by **Greg** » Thu Mar 13, 2008 7:40 pm

Tony Moore wrote:...perhaps they all think I'm mad?

Ah yes, I told them all about the 'Madbox' capacitor bank

Post by **Nick** » Thu Mar 13, 2008 8:41 pm

Hey, these are people who stack 16 DACs

Andrew · Post by **Andrew** » Thu Mar 13, 2008 9:21 pm

SimonC wrote:
Maybe we should be casting our nets further afield and look into some of the new chips on the market?

Simon C

Might be a thought, and a very good one, but I doubt you'd find any NOS chips, tho' that might be OK as long as its the agreed goal. Are we attached to NOS?

[edit] To add to this, its clear NOS without a brick-wall filter measures off the norm, what worries me is Nick's aural impressions and I trust his ears.

cheers,

-- Andrew

iansr · Post by **iansr** » Thu Mar 13, 2008 9:25 pm

Just caught up with this thread. I'm happy to bring along the aforementioned Cambridge DAC to Owsten if people are interested in a DAC comparison. Alternatively, I could bring my Sugden CDP which has a 1541 S2 (Double Crown). For the uninitiated, these were the cream of the 1541 crop - perfect examples allegedly capable of full 16 bit conversion. This CDP has digital input and output connectors so we could use it for comparisons.

On the subject of newer DACs, AKM boast that their latest DACs combine the advantages both ladder and sigma delta types whilst overcoming the disadvantages of each. I can't comment furhter but at the very least they've developed a good line in hype

Post by **Nick** » Thu Mar 13, 2008 10:49 pm

Reading that (very interesting) link Ed posted has got me to think that there are at least three variables,

1. Over Sample or not
2. DAC type, ladder or sigma delta
3. Reconstruction filter

And I think each of those can be split further.I think one problem with enewer DAC chips is its diffucult to split 1 and 2.

Ian: I think it would be great if you could bring at least the dac to Owlston, I would certainly like another chance to hear it (and also have another look inside if possible to see just whats in there).

Andrew: my gut feeling at the moment is that current DACs may be going in the wrong direction, with our nyquest hats on, we both know that vinyl is actually less than 14 bit resolution, has some major HF issues, and has mono bass, but that doesn't seem to stop it sounding very very good.

audio-talk

DAC designs

#16

#17

#18

#19

#20

#21

#22

#23

#24

#25

#26

#27

#28

#29

#30