PM from hollowman
Quote> Guido Tent just repled to my query about which *possible" XO clock he may have been referring to -- i.e., the one he highly recommended <http://www.diyaudio.com/forums/showthread.php?postid=1185658#post1185658>. His reply is not that *definite*.
>
> I also noted that you, at one point tried, one of the Tent clocks , <Quote
I have not yet played with or instaled an XODAC clock for dac's.
Only the XO3 & XO2 and the XO power supplies for transports.
Below is the front end of the A3.24, and yes it has a 50mhz clock standard, maybe Guido changes this to 24mhz when he used his XODAC, in the one he modded, maybe he will answer you.
Cheers George
Quote> Guido Tent just repled to my query about which *possible" XO clock he may have been referring to -- i.e., the one he highly recommended <http://www.diyaudio.com/forums/showthread.php?postid=1185658#post1185658>. His reply is not that *definite*.
>
> I also noted that you, at one point tried, one of the Tent clocks , <Quote
I have not yet played with or instaled an XODAC clock for dac's.
Only the XO3 & XO2 and the XO power supplies for transports.
Below is the front end of the A3.24, and yes it has a 50mhz clock standard, maybe Guido changes this to 24mhz when he used his XODAC, in the one he modded, maybe he will answer you.
Cheers George
Attachments
Georgehifi: Thx for pursuing my query!
Honestly, though, I'm a bit confused: the seemingly UNscientific method in which some DIYers/developers swap various-Mhz oscillators and anecdotally report one sounding better than another is bit like practicing the black arts. E.g.:
http://diyparadise.com/yabb/YaBB.pl?num=1168131286/5#5
http://diyparadise.com/yabb/YaBB.pl?num=1168131286/7#7
Question:
What is/are a good SCIENTIFIC (=repeatable) method(s) for selecting the best (frequency) oscillator to pair with a particular stand-alone DAC (i.e., using asynchronous reclocking)? If one is tweaking/modding a pre-manufactured unit, then one may just look at the frequency (MHz) rating on the old oscillator (provided that the manuf. used sound scientific and/or empirical-listening tests as part of their selection criteria). But what if one is building a DAC from scratch?
Tent labs *claim* their oscillators have low jitter, so that's mostly why I'm interested in their product**.
I'm planning on pairing a Tent oscillator with, possibly, a Kwak-Clock (found elsewhere on this forum). For clean power, I may go with one of these routes. The kit modules from Tent, Audiocom, etc. are way too rich for my blood.
**Anyone know where "audiophile" companies -- e.g. Tent, LCAudio, Audiocom -- source their oscillators from. I suspect it may be a company like Crystek. This company makes low-jitter oscillators, but their thru-hole stuff do not seem to be avail. for end users (Mouser does carry most of their SMD stuff).
Guido Tent said:
But this is not what I meant by "black art". Tight manuf. specs and "measured perfomance", as on a test bench, is exacting science. However, the scope here is narrow: it does not conote better sound.
Rather (and repeating from message above), "black art" is meant to address the following query:
Scientific validation could be made rigorously efficacious:
If a company that makes audiophile clocks (or other tweaks for that matter) has promoted the use of empirically-siginificant metrical/specs tests (jitter, THD, linearity, etc.) on a broad range of modified units (players, outboard DACs), conducted by an independent organization (e.g. ISO), then that's science.
And/or:
If a company that makes audiophile clocks (or other tweaks for that matter) has promoted the use of empirically-siginificant listening tests (incl. subjective ones) on broad range of modified units (players, outboard DACs), conducted by an independent organization (e.g. ISO), then that's science.
Scientific validation could also be gleaned...
If there was/is popular consensus (= empirically statistically valid) among a large group of people (e.g., diyAudio forum members) that a certain product has/had certain qualities (sounds "good"/"bad").
hollowman said:
Unlike with distortion, or bandwidth or amplitude flatness, with jiter it is easy: Lower jitter means better quality.
That is, if you are after transpararent reproduction. If you're after something else, you'd need different types of jitter. i can make these, but won't sell them.
cheers
rfbrw said:
Your assessment does not reflect de facto standard:
The majority of commercial and DIY stand-alone DACs I've encountered contain oscillators.
As far as ASRC being the wrong "geometry"... you may be right. While I haven't heard them all, I generally don't like the sound of NOS DACs, many of which (TTBOMK) use ASRC.
That said, Doede Douma's NOS ASRC parallel-processor <http://dddac.de/> is quite popular among Internet DIYers.
Alas, there seems to be no easy answer to my orig. query.
hollowman said:
...and/or the clock freq. isn't all that important. I.e. there are other factors -- circuit topology, innate low-jitter of oscillator, clean power to clock circuit, etc. -- that weigh just as heavily as (if not more than) freq. This is sheer speculation, though ... and will remain that way until those more topically knowledgeable weigh in...
hollowman said:
One can only assume you have never read, or if read failed to understand, a DAIR datasheet. The oscillator is not necessary. It is there to provide a signal for the downstream IC's when the SPDIF datastream is not present. That way, the dac effectively converts digital silence instead of producing noise. The MCLK signal used during playback is derived from the SPDIF datastream.
Asynchronous reclocking and ASRC are two different concepts. Originally intended the use of different sample rates e.g. converting from 44K1 to 48k in a broadcast arena without huge expense of multiple DSP chips, ASRC's (Asynchronous Sample Rate Converter) are also highly effective jitter suppressors.
The MF A3.24 uses two. The first converts all incoming data to 96k and at the same time suppresses jitter by creating a new clock domain hence the need for an oscillator. The clock domain associated with the clock embedded in the SPDIF datastream is not used. The second ASRC provides an optional increase to 192K.
Asynchronous reclocking, OTOH, is a daft idea that adds jitter. The is no science involved anymore than there is in deciding how best to poke yourself in the eye with a sharp stick. The dddac does not use an ASRC. Instead it allows for the option of asynchronous reclocking.
Only because you are asking the wrong question.
rfbrw said:
The jitter-supprssion effect is noteworthy. As I understood this from the AD1896 datasheet, it upconverts digital signals into an ultra-high-speed, "high-resolution" digital signal ditching the incoming clock's data, and then downconverts it into the slower 192kHz word stream while interpolating (to get rid of the "digital" haze) and reclocking (to get rid of jitter) before sending the data off to the DACs.
This processing causes some "hold time" of the data. So I wonder if a "simpler" mechanism isn't in effect (i.e., responsible for jitter-reduction) -- something similar to what's going on in this now-ancient jitter-reduction tool:
http://www.stereophile.com/digitalprocessors/824/index1.html
Actually, I may have kinda/sorta answered my own question (see above). But If I didn't, what's the "right" question?
rfbrw said:
You correctly deduced that I was, in fact, erroneously equating ASRC with "asynchronous reclocking". So thanks for the correction and detailed follow-up.
Fascinated by the topic, I did a bit of digging, albeit non-exhaustive, and found that there is no clear-cut consensus for or against asynchronous reclocking.
FWIW, the developers at LessLoss.com <http://www.lessloss.com/faq.html> note:
"If you use asynchronous reclocking, you are continuously rounding out the digital audio stream to match the new clock. There is a threshold of incoming Jitter amount or type from where this is a good feature (it sounds better with asynchronous reclocking rather than without). But if your incoming Jitter level is LESS than that threshold, then the asychronous mathematical rounding process actually makes the audio worse, since there is a new type of distortion entering the data stream (this is not necessarily measureable Jitter, but the effect of these asynchronous processes on the original audio data stream actually encode a new digital stream with rounding errors)."
Also on this page <http://www.lessloss.com/types.html>, the developers do not feel async reclocking is the worst of their tested strategies. E.g. "Digital Slave Mode, 8x Oversampling, No Reclocking" is worse than "Digital Slave Mode, 8x Oversampling, Asynchronous Reclocking". This developer has, of course, a financial interest in their product.
Some builders of DDDAC also claim positive benefits:
http://www.dddac.de/ma_dac25a.htm
I don't think this is correct for two reasons:
(1) The main page for DDDAC1543 MK2 <http://dddac.de/ma_dac31.htm> notes:
"Non-Oversampling, I2S-BUS, Re-Clocking Modular DAC system with SPDIF and USB Input...." No where do I see this as an "option". But one could certainly build the DAC w/o asynchronous reclocking (see (2)). Asynchronous reclocking is default and not deliberately offered as an option. Im guessing the developer believes it is overall a good thing to have.
(2) I do know of at least one builder of DDDAC who did not care for the sound of the default asynchronous reclocking, and chose to remove the feature from his DDDAC. Specifically, he noted to me a few months back: "I realized it was more or less degrading the sound once you have a good transport source. I don't reclock any longer." His former expereince with reclocking was positive <http://didnt.doit.wisc.edu/audio/dddac1543/dddac1543.html>.
Conclusion:
If you have an opinionated argument to make on this topic, then please use a disclaimer (e.g. "IMO") and/or back up your opinion with references/links. Otherwise, you're putting words into others' mouths.
P.S.
Apologies in advance for not containing all the replies in one message. Had to do some fact-checking, first. But all reading this post can decide/comment-on on the overall accuracy.
Correction on ASR; more digging
As soon as I posted this message, I found that an older version of DDDAC that does indeed have a switch for this "option":
http://www.dddac.de/ma_dac21.htm
and is also noted in this thread message:
http://www.diyaudio.com/forums/showthread.php?postid=334738#post334738
So my apologies to rfbrw and readers for jumping the gun here!
BTW, the parent thread for the above link...
http://www.diyaudio.com/forums/showthread.php?postid=334738
... reaffirms my earlier position (opinion) on ASR in that there is no solid consensus on this issue.
hollowman said:
As soon as I posted this message, I found that an older version of DDDAC that does indeed have a switch for this "option":
http://www.dddac.de/ma_dac21.htm
and is also noted in this thread message:
http://www.diyaudio.com/forums/showthread.php?postid=334738#post334738
So my apologies to rfbrw and readers for jumping the gun here!
BTW, the parent thread for the above link...
http://www.diyaudio.com/forums/showthread.php?postid=334738
... reaffirms my earlier position (opinion) on ASR in that there is no solid consensus on this issue.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.